WO2023024542A1 - 车辆决策规划方法、装置、设备及介质 - Google Patents
车辆决策规划方法、装置、设备及介质 Download PDFInfo
- Publication number
- WO2023024542A1 WO2023024542A1 PCT/CN2022/088532 CN2022088532W WO2023024542A1 WO 2023024542 A1 WO2023024542 A1 WO 2023024542A1 CN 2022088532 W CN2022088532 W CN 2022088532W WO 2023024542 A1 WO2023024542 A1 WO 2023024542A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- obstacle
- information
- grid
- lane
- vehicle
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 75
- 230000003068 static effect Effects 0.000 claims description 67
- 230000008447 perception Effects 0.000 claims description 55
- 238000007781 pre-processing Methods 0.000 claims description 18
- 238000001914 filtration Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 12
- 230000002776 aggregation Effects 0.000 claims description 10
- 238000004220 aggregation Methods 0.000 claims description 10
- 238000003860 storage Methods 0.000 claims description 9
- 238000004891 communication Methods 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 abstract description 10
- 230000000875 corresponding effect Effects 0.000 description 67
- 239000013598 vector Substances 0.000 description 44
- 238000010586 diagram Methods 0.000 description 31
- 238000004422 calculation algorithm Methods 0.000 description 20
- 238000001514 detection method Methods 0.000 description 18
- 230000007613 environmental effect Effects 0.000 description 14
- 238000004590 computer program Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 5
- 230000002596 correlated effect Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000005457 optimization Methods 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 230000004931 aggregating effect Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005266 casting Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000009966 trimming Methods 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W30/00—Purposes of road vehicle drive control systems not related to the control of a particular sub-unit, e.g. of systems using conjoint control of vehicle sub-units
- B60W30/08—Active safety systems predicting or avoiding probable or impending collision or attempting to minimise its consequences
- B60W30/095—Predicting travel path or likelihood of collision
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W30/00—Purposes of road vehicle drive control systems not related to the control of a particular sub-unit, e.g. of systems using conjoint control of vehicle sub-units
- B60W30/08—Active safety systems predicting or avoiding probable or impending collision or attempting to minimise its consequences
- B60W30/095—Predicting travel path or likelihood of collision
- B60W30/0956—Predicting travel path or likelihood of collision the prediction being responsive to traffic or environmental parameters
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W30/00—Purposes of road vehicle drive control systems not related to the control of a particular sub-unit, e.g. of systems using conjoint control of vehicle sub-units
- B60W30/08—Active safety systems predicting or avoiding probable or impending collision or attempting to minimise its consequences
- B60W30/09—Taking automatic action to avoid collision, e.g. braking and steering
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W30/00—Purposes of road vehicle drive control systems not related to the control of a particular sub-unit, e.g. of systems using conjoint control of vehicle sub-units
- B60W30/08—Active safety systems predicting or avoiding probable or impending collision or attempting to minimise its consequences
- B60W30/085—Taking automatic action to adjust vehicle attitude in preparation for collision, e.g. braking for nose dropping
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/02—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to ambient conditions
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
- B60W50/0098—Details of control systems ensuring comfort, safety or stability not otherwise provided for
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W60/00—Drive control systems specially adapted for autonomous road vehicles
- B60W60/001—Planning or execution of driving tasks
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/16—Anti-collision systems
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
- B60W2050/0001—Details of the control system
- B60W2050/0043—Signal treatments, identification of variables or parameters, parameter estimation or state estimation
- B60W2050/0052—Filtering, filters
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2552/00—Input parameters relating to infrastructure
- B60W2552/30—Road curve radius
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2552/00—Input parameters relating to infrastructure
- B60W2552/53—Road markings, e.g. lane marker or crosswalk
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2554/00—Input parameters relating to objects
- B60W2554/20—Static objects
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2554/00—Input parameters relating to objects
- B60W2554/40—Dynamic objects, e.g. animals, windblown objects
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2556/00—Input parameters relating to data
- B60W2556/40—High definition maps
Definitions
- the present disclosure relates to the technical field of unmanned driving, and in particular to a vehicle decision planning method, device, equipment and medium.
- the automatic driving system needs to plan a smooth, safe and vehicle-passable path to ensure that the vehicle will not collide with obstacles.
- the perception module of the automatic driving system will output two types of obstacles, one is a convex hull obstacle with rich semantic information, and the other is a grid obstacle without semantic information.
- the decision planning module can make obstacle decisions more conveniently, but for grid obstacles with high discreteness and a large number, it is difficult for the decision planning module to make obstacle decisions conveniently and quickly , which makes it difficult for the decision planning module to make obstacle decisions for mixed types of obstacles.
- the present disclosure provides a vehicle decision planning method, device, equipment and medium.
- An embodiment of the present disclosure provides a vehicle decision-making planning method, including:
- obstacle decisions are made, including:
- the target convex-hull obstacle includes the first convex-hull obstacle and/or the The second convex hull obstacle.
- the drivable area is generated based on obstacle decisions, including:
- the environment awareness information includes at least two of lane information, obstacle information and vehicle information, and the obstacle information includes static obstacle information and/or dynamic obstacle information;
- lane decision semantic information Based on the environment perception information, determine lane decision semantic information of each lane, wherein the lane decision semantic information includes passing time cost and safety cost;
- a drivable area is generated.
- An embodiment of the present disclosure provides a vehicle decision planning device, including:
- the base coordinate system generator is used to generate the base coordinate system
- a guideline generator configured to generate a guideline under the base coordinate system to determine a rough future trajectory of the vehicle
- An obstacle decision maker configured to perform obstacle decision-making under the constraints of the guide line
- Driving space generator for generating drivable areas based on obstacle decisions.
- the obstacle decider includes:
- An information acquisition module configured to acquire road information, first grid obstacle information of the first grid obstacle, and first convex hull obstacle information of the first convex hull obstacle;
- a preprocessing module configured to preprocess the first grid obstacle based on the road information and the first grid obstacle information to obtain a second grid obstacle, wherein the second grid obstacle The number of grid obstacles is less than the number of the first grid obstacles;
- a type conversion module configured to convert the second grid obstacle into a second convex hull obstacle
- An avoidance decision-making module configured to make an avoidance decision for the target convex-hull obstacle based on the target convex-hull obstacle information of the target convex-hull obstacle, wherein the target convex-hull obstacle includes the first convex-hull obstacle object and/or the second convex obstacle.
- the drive space generator includes:
- a perception information acquisition module configured to acquire environment perception information, wherein the environment perception information includes at least two of lane information, obstacle information and vehicle information, and the obstacle information includes static obstacle information and/or dynamic Obstacle information;
- a lane decision semantic information determination module configured to determine the lane decision semantic information of each lane based on the environment perception information, wherein the lane decision semantic information includes passing time cost and safety cost;
- a drivable area generating module configured to generate a drivable area based on the lane decision semantic information.
- An embodiment of the present disclosure provides an electronic device, including:
- the memory is connected in communication with the one or more processors, the memory stores instructions executable by the one or more processors, and the instructions are executed by the one or more processors , the electronic device is used to implement the vehicle decision planning method provided in any embodiment of the present disclosure.
- An embodiment of the present disclosure provides a computer-readable storage medium on which computer-executable instructions are stored.
- the computer-executable instructions When executed by a computing device, they can be used to implement the vehicle decision-making plan provided by any embodiment of the present disclosure. method.
- FIG. 1 is a schematic flowchart of a vehicle decision planning method provided by an embodiment of the present disclosure
- FIG. 2 is a functional block diagram of a decision planning module provided by an embodiment of the present disclosure
- FIG. 3 is a schematic diagram of an applicable scenario of a vehicle decision planning method provided by an embodiment of the present disclosure
- FIG. 4 is a schematic flowchart of a part of a vehicle decision-making planning method provided by an embodiment of the present disclosure
- FIG. 5 is a functional module block diagram of an obstacle decision maker provided by an embodiment of the present disclosure.
- FIG. 6 is a schematic diagram of a road bounding box provided by an embodiment of the present disclosure.
- FIG. 7 is a scene diagram of an avoidance decision provided by an embodiment of the present disclosure.
- FIG. 8 is a partial flowchart of another vehicle decision planning method provided by an embodiment of the present disclosure.
- FIG. 9 is a scene diagram corresponding to the lane passing time cost provided by the embodiment of the present disclosure.
- FIG. 10 is an ST diagram of lane safety judgment provided by an embodiment of the present disclosure.
- FIG. 11 is a schematic diagram of the discretization of the boundary of the drivable area provided by the embodiment of the present disclosure.
- FIG. 12 is a scene diagram corresponding to the lane width traffic cost provided by the embodiment of the present disclosure.
- FIG. 13 is a schematic diagram of updating a drivable area based on preset traffic rules provided by an embodiment of the present disclosure
- FIG. 14 is a schematic diagram of updating a drivable area with kinematics and dynamics constraints of a vehicle provided by an embodiment of the present disclosure
- Fig. 15 is a schematic diagram of an obstacle semantic information and a preset safe area update drivable area provided by an embodiment of the present disclosure
- FIG. 16 is a schematic diagram of generating a frenet bounding box provided by an embodiment of the present disclosure.
- Fig. 17 is a schematic diagram of static obstacle information, obstacle decision semantic information and ray tracing algorithm updating the drivable area provided by an embodiment of the present disclosure
- FIG. 18 is a block diagram of functional modules of an obstacle decision maker in a vehicle decision planning device provided by an embodiment of the present disclosure
- FIG. 19 is a block diagram of functional modules of the driving space generator in the vehicle decision planning device provided by an embodiment of the present disclosure.
- FIG. 20 is a schematic structural diagram of an electronic device suitable for realizing the implementation manners of the present disclosure provided by the embodiments of the present disclosure.
- Fig. 1 is a schematic flowchart of a vehicle decision planning method provided by an embodiment of the present disclosure. This method is suitable for obstacle decision-making and drivable area generation of unmanned vehicles. As shown in Figure 1, the method includes the following steps:
- the base coordinate system may be a frenet coordinate system.
- obstacle decision-making is made on the obstacle on the approximate travel trajectory, so as to avoid the obstacle.
- the obstacle decision it can be determined whether the vehicle needs to pass the obstacle on the left side, pass the obstacle on the right side, or follow the obstacle, so as to determine the driving area of the vehicle.
- Fig. 2 shows a block diagram of the functional modules of the decision planning module.
- the decision planning module 1 may include a constraint generation unit 11 , a trajectory generation unit 12 and a trajectory smoothing unit 13 .
- the constraint generating unit 11 includes a base coordinate system generator 111 , a guiding line generator 112 , an obstacle decision maker 113 and a driving space generator 114 .
- the base coordinate system generator 111 is used to generate a base coordinate system, such as a frenet coordinate system, etc.
- the guideline generator 112 is used to generate a guideline to determine a general trajectory of the vehicle in the future
- the obstacle decision maker 113 is used to perform Obstacle Decisions
- Driving Space Generator is used to generate drivable areas based on obstacle decisions.
- the trajectory generation unit 12 is used to generate the driving trajectory of the unmanned vehicle according to the drivable area
- the trajectory smoothing unit 13 is used to smooth the driving trajectory.
- the obstacle decision maker 113 is specifically configured to obtain road information, the first grid obstacle information of the first grid obstacle, and the first convex hull obstacle information of the first convex hull obstacle; based on the road information and the first grid obstacle information, preprocess the first grid obstacle to obtain the second grid obstacle; convert the second grid obstacle into the second convex hull obstacle; based on the target convex hull obstacle
- the target convex hull obstacle information of the object is used to make an avoidance decision for the target convex hull obstacle.
- the driving space generator 114 is specifically used to obtain environmental awareness information; based on the environmental awareness information, determine the lane decision semantic information of each lane, wherein the lane decision semantic information includes passing time cost and safety cost; based on Lane decision semantic information to generate drivable areas.
- the embodiment of the present disclosure provides a vehicle decision-making planning method, which is suitable for unmanned vehicles to detect static obstacles and/or dynamic obstacles such as grid obstacles and convex hull obstacles in the road environment.
- Fig. 3 shows the applicable scenario of the vehicle decision-making planning method.
- convex hull obstacles 200 including static obstacles and dynamic obstacles
- grid obstacles 300 in front of the unmanned vehicle 100.
- the car 100 can convert the type of the grid obstacle 300 into a convex hull type by obtaining the obstacle information of the convex hull obstacle 200 and the grid obstacle 300, so as to realize the unification of the convex hull obstacle 200 and the grid obstacle 300 decision making.
- This method can be applied to unmanned vehicles, and is specifically applied to the decision planning module in the automatic driving system of unmanned vehicles. Based on the vehicle decision-making planning method provided by the embodiments of the present disclosure, unified decision-making for mixed-type obstacles can be realized, and obstacle decision-making can be made conveniently and quickly.
- FIG. 4 is a partial flowchart of a vehicle decision planning method provided by an embodiment of the present disclosure. As shown in Figure 4, making an obstacle decision (or a decision-making method for avoiding an obstacle) includes the following steps:
- the grid obstacle is a grid type obstacle
- the convex hull obstacle is a convex hull type obstacle
- the road information can be acquired through a high-precision map or a vehicle-mounted camera, and the road information can include road boundary information and road curvature information.
- obstacle information can be obtained through vehicle perception modules (such as on-board cameras and laser radars, etc.) and positioning modules. Obstacle information can include obstacle type information, obstacle size information, and obstacle location information.
- the obstacle type information can be an obstacle type identifier, and the obstacle type is distinguished by pre-defining different obstacle type identifiers; the obstacle type information can also be an obstacle data format, and the vehicle perception module perceives the obstacle
- the obstacle data is processed, and the data of different types of obstacles are stored in different obstacle data formats, so that the decision planning module can distinguish obstacle types through obstacle data formats when obtaining obstacle information, for example,
- the obstacle data format of grid obstacles is ".ogm”
- the obstacle data format of convex hull obstacles is ".mot”.
- the first grid obstacle and the first convex hull obstacle are determined to obtain the first grid obstacle information and the first convex hull obstacle information.
- Fig. 5 shows a block diagram of the functional modules of the obstacle decision maker.
- the obstacle decision maker 113 may include a grid obstacle processor 1131 , a way of passage decision maker 1132 and a convex hull obstacle filter 1133 .
- the grid obstacle processor 1131 executes S220, based on the road information and the first grid obstacle information, preprocessing the first grid obstacle to obtain the second grid obstacle; and S230, the second grid obstacle information Grid obstacles are transformed into second convex hull obstacles.
- preprocessing the first grid obstacle can be used to reduce the amount of data calculation and simplify the obstacle decision-making process, which may include at least one of the following steps: generating the grid obstacle outline of the first grid obstacle; generating the second grid obstacle An obstacle bounding box of a grid obstacle; filtering out the first grid obstacle located outside the road; and performing aggregation processing on the first grid obstacle located inside the road.
- preprocessing is performed on the first grid obstacles, so that the number of second grid obstacles obtained after preprocessing is smaller than the number of first grid obstacles, so that the downstream modules can easily detect the obstacles calculate.
- Embodiments of the present disclosure may reduce the number of first grid obstacles by filtering out first grid obstacles located outside the road.
- preprocessing the first grid obstacle to obtain the second grid obstacle based on the road information and the first grid obstacle information may include the following steps:
- filtering out the first grid obstacle located outside the road may include the following steps:
- the road boundary is discretized into boundary points; based on the boundary points, a road bounding box is generated.
- the shape of the road bounding box is based on the axisymmetric bounding box of the right-hand coordinate system of the unmanned vehicle body, as long as it can pass through the boundary point and cover the road, so that the subsequent determination of whether the first grid obstacle is located between the road outside judgment.
- the road boundary and the road boundary curvature are determined; based on the road boundary curvature, the road boundary is discretized to obtain boundary point groups arranged at intervals along the road traffic direction, wherein each group of boundary points The group includes the corresponding left boundary point a and right boundary point a' in the transverse direction, and the transverse direction is perpendicular to the road traffic direction; based on any adjacent two groups of boundary point groups, generate the Rectangular box b, and use rectangular box b as the road bounding box.
- one side is parallel to the driving direction of the self-driving car, that is, the driving direction x of the unmanned vehicle 100, and the other side is perpendicular to the driving direction of the self-driving car, that is, no one is driving.
- the normal direction y in which the car 100 travels is driven.
- the distance between two adjacent boundary points on the same road boundary is negatively correlated with the road boundary curvature, that is, the greater the curvature of the road boundary, the greater the degree of curvature, and the distance between two adjacent boundary points on the road boundary smaller.
- the road boundary is discretized to obtain a group of boundary points arranged at intervals along the road traffic direction, including: taking the current position of the vehicle as the starting waypoint; obtaining the starting waypoint at A group of corresponding boundary points in the horizontal direction; based on the curvature of the road boundary, select the next road point along the direction of road traffic, where the distance between two adjacent road points is negatively correlated with the curvature of the road boundary; the next road point is used as the starting point Waypoint, return to execute to obtain a set of boundary point groups corresponding to the initial waypoint in the horizontal direction, until the distance from the up and down waypoints to the current position of the vehicle in the direction of road traffic is greater than the preset distance threshold, all currently obtained boundary point groups identified as boundary point groups.
- the preset distance threshold may be determined according to the maximum range in which the vehicle perceives obstacles.
- a road bounding box B R ⁇ b min ,b max ,b left,0 ,b left,1 ,b right,0 ,b right,1 ⁇
- b min and b max are the minimum and maximum coordinate points of the road bounding box
- b left, 0 ,b left,1 ,b right,0 ,b right,1 are the coordinate points on the left and right sides of the road respectively
- the whole road can be represented by n road bounding boxes
- the road bounding box sequence B road_list ⁇ B R0 ,B R1 ,L,B Rn ⁇ .
- the road may be a route segment in the driving route of the vehicle, and the route segment where the vehicle is located can be determined according to the vehicle positioning information, and the road boundary is discretized, that is, the boundary of the route segment where the vehicle is located is discretized change.
- the list of road boundary points is defined as S, and it is initialized as an empty table, and the road bounding box sequence B road_list is also initialized as empty, and the road boundary is discretized from the route segment where the vehicle is located, and the first road of the route segment is obtained point (it can be the current position of the vehicle), then obtain the left boundary point and right boundary point corresponding to the first waypoint in the horizontal direction, and add the current left boundary point and right boundary point to the list S; based on the road boundary curvature, Select the next waypoint along the direction of the road, check whether the distance from the first waypoint to the next waypoint along the direction of the road is less than or equal to the preset distance threshold, if the distance is less than or equal to the preset distance threshold, then get the next waypoint The corresponding left boundary point and right boundary point in the horizontal direction are added to the list S; based on the curvature of the road boundary, continue to select the next waypoint along the road traffic direction until the first waypoint is along the road traffic direction to the next way
- a grid obstacle contour of the first grid obstacle is generated; and a grid obstacle bounding box is generated based on the grid obstacle contour.
- the suzuki contour tracking algorithm is used to generate a closed contour figure of the obstacle in the first grid, that is, the contour of the obstacle in the grid.
- the grid obstacle profile consists of n coordinate points.
- x 0 , x 1 ,..., x n are the x coordinates of n coordinate points in the grid obstacle outline
- y 0 , y 1 ,..., y n are the y coordinates of n coordinate points in the grid obstacle outline.
- the embodiment of the present disclosure can perform two-stage collision detection on the first grid obstacle, so as to quickly and accurately determine the first grid obstacle located outside the road.
- rough collision detection can be performed on grid obstacles first, so as to quickly filter out the first grid obstacle located outside the road, and reduce the calculation amount of collision detection; for the first collision detected by rough collision detection One grid obstacle, then perform fine collision detection, so as to further determine the first grid obstacle located outside the road, so as to ensure that the remaining first grid obstacles after filtering are all located inside the road.
- the Euclidean distance to the grid obstacle bounding box is determined from the road bounding box The smallest target road bounding box; perform collision detection between the grid obstacle bounding box and the corresponding target road bounding box; if there is no collision between the grid obstacle bounding box and the corresponding target road bounding box, determine the grid obstacle bounding box The corresponding first grid obstacle is located outside the road.
- the Euclidean distance from the road bounding box to the grid obstacle bounding box is relatively small, it indicates that the collision between the road bounding box and the grid obstacle bounding box is more likely.
- the calculation amount of collision detection can be reduced, thereby speeding up obstacle decision-making speed. In some embodiments, it is only necessary to detect whether the vertices of the obstacle bounding box of the grid are located on or within the bounding box of the target road.
- the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road; when the grid obstacle bounding box When the vertex is located on or within the target road bounding box, it is determined that the first grid obstacle corresponding to the grid obstacle bounding box is located within the road.
- collision detection is performed based on the boundary points of the target road bounding box and the grid obstacle bounding box by vector cross product, and it is determined whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road. Specifically, determine the boundary point vector; determine the vertex vector of the grid obstacle bounding box; when the cross product of the vertex vector of the grid obstacle bounding box and the boundary point vector is greater than 0, determine the grid obstacle bounding box corresponding The first grid obstacle is located outside the road.
- the boundary point vector includes the left boundary vector formed by the two left boundary points of the target road bounding box, and the right boundary vector formed by the two right boundary points of the target road bounding box, and the vertex vector of the grid obstacle bounding box is grid
- the vector formed by the vertex of the grid obstacle bounding box and a boundary point of the target road bounding box, and the boundary point is a boundary point corresponding to the boundary point vector participating in the cross product operation, for example, when the vertex vector cross-products with the right boundary vector , the boundary point in the vertex vector is a boundary point corresponding to the right boundary vector.
- grid obstacle bounding box B ⁇ P min ,P max ⁇
- target road bounding box B R ⁇ b min ,b max ,b left,0 ,b left,1 ,b right,0 ,b right,1 ⁇
- the left boundary vector v left b left,1 -b left,0
- it can be determined that other vertices of the grid obstacle bounding box are on the right boundary of the road or on the left or right of the right boundary of the road. In this way, it can be determined whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.
- the embodiments of the present disclosure may also reduce the number of first grid obstacles located in the road by performing aggregation processing on the first grid obstacles located in the road.
- preprocessing the first grid obstacle to obtain the second grid obstacle based on the road information and the first grid obstacle information may also include the following steps:
- the first grid obstacle located inside the road can be determined by the method of judging whether the first grid obstacle is located outside the road in the above embodiment, and details will not be repeated here.
- the first obstacle bounding box of the first grid obstacle is generated; when two adjacent first obstacles When the Euclidean distance between object bounding boxes is less than the width of the vehicle, merge two adjacent first obstacle bounding boxes to generate a second obstacle bounding box; use the second obstacle bounding box as the first obstacle surround Box, return to execute when the Euclidean distance between two adjacent first obstacle bounding boxes is less than the vehicle width, merge two adjacent first obstacle bounding boxes to generate a second obstacle bounding box, until The Euclidean distance between the second obstacle bounding box and the adjacent first obstacle bounding box is greater than or equal to the vehicle width, or there is no first obstacle bounding box adjacent to the second obstacle bounding box.
- a CLOSED table can be created and initialized as an empty table, a first obstacle bounding box is taken from the set contour of the first obstacle bounding box and added to the CLOSED table, and in the set set contour Delete the first obstacle bounding box, and then traverse the set set contour , once the Euclidean distance between the first obstacle bounding box in the set set contour and the first obstacle bounding box in the CLOSED table is less than the width of the vehicle, set The first obstacle bounding box of set contour is added to the CLOSED table, and aggregated with the first obstacle encircled by the Euclidean distance comparison in the CLOSED table, aggregated into a new first obstacle encircled, and then added to the CLOSED table The first obstacle bounding box is removed from the set set contour . This cycle repeats until the set set contour is empty, and then the aggregation processing of the first grid obstacle located in the road can be completed.
- the embodiment of the present disclosure can also filter out the first grid obstacle located outside the road based on the road information and the first grid obstacle information, and determine The first grid obstacle located in the road, and aggregation processing is performed on the first grid obstacle located in the road. In this way, the number of first grid obstacles can be further reduced.
- a fast convex hull algorithm may be used to convert the second grid obstacle into a second convex hull obstacle. In this way, unified decision-making for grid obstacles and convex hull obstacles can be realized.
- This step can be executed by the traffic mode decider 1132 in FIG. 5 .
- making an avoidance decision for the target convex hull obstacle includes: based on the target convex hull obstacle information, for the target meeting the preset filter condition Convex obstacles are marked with no avoidance labels or no lateral avoidance labels.
- the trajectory generation unit 12 in FIG. 2 can ignore the target convex-hull obstacle, which can reduce trajectory generation.
- the burden of obstacles processed by the unit 12 can increase the trajectory generation speed and improve the rationality of trajectory generation.
- the preset filtering conditions include at least one of the following: the target convex-hull obstacle is located outside the road; the motion state of the target convex-hull obstacle satisfies the condition of no need for lateral avoidance; the target convex-hull obstacle is located on the leading line of the vehicle superior.
- the target convex-hull obstacle information label the target convex-hull obstacle that satisfies the preset filtering conditions with no avoidance label or no lateral avoidance label, including: when the target convex-hull obstacle is outside the road, mark the target convex-hull obstacle Label the obstacle without avoidance; when the movement state of the obstacle with convex hull meets the condition of avoidance without lateral avoidance or the obstacle with convex hull is located on the leading line of the vehicle, label the obstacle with avoidance without lateral avoidance.
- the target convex hull obstacle when the target convex hull obstacle is located outside the road, such as obstacle 1, the target convex hull obstacle has no influence on the normal driving of the driverless car 100 at all, and the target convex hull obstacle can be ignored at this time. package obstacles, and label the target convex hull obstacle without avoidance.
- the driverless car 100 When the motion state of the target convex-hull obstacle satisfies the condition of no need for lateral avoidance, for example, when the target convex-hull obstacle crosses the road, such as a pedestrian crossing the road, the driverless car 100 only needs to wait for the pedestrian to pass, without generating a The trajectory of the detour can be marked on the pedestrian without a lateral avoidance label; another example is that the target convex obstacle changes lanes to the vehicle lane, or the longitudinal speed of the target convex obstacle is greater than the vehicle speed (such as adjacent lanes moving at high speed) Obstacles), without affecting the lane safety of the vehicle, the vehicle does not need to avoid laterally, and the above-mentioned target convex-hull obstacle can be marked with no need to avoid laterally; if the target convex-hull obstacle is located on the guidance line of the vehicle , if the obstacle 2 is located on the leading line c of the vehicle, the unmanned vehicle 100 does not need to avoid the obstacle 2 laterally,
- the convex hull obstacle satisfying the preset filter condition can be filtered out through the convex hull obstacle filter 1133 in FIG. 5 .
- the convex hull obstacle filter 1133 may include at least one of a road network filter based on obstacle frenet bounding boxes, a behavior semantic information filter, and a guide line filter.
- the road network filter based on the obstacle frenet bounding box can quickly filter out the target convex hull obstacle located outside the road, and the road network filter based on the obstacle frenet bounding box can adopt the two-stage collision in the above embodiment Detection method to filter out the target convex hull obstacles located outside the road;
- the behavioral semantic information filter can filter out the target convex hull obstacles that do not need to avoid and do not need to avoid laterally according to the semantic information contained in the target convex hull obstacles; guide lines
- the filter filters out the target convex hull obstacles that collide with the guideline.
- the embodiments of the present disclosure can also Make an avoidance decision of following, passing on the left or passing on the right.
- marking the target convex hull obstacle with an avoidance label may also include: if the target convex hull obstacle is located on the vehicle guidance line, labeling the target convex hull obstacle If the target convex-hull obstacle is not located on the leading line of the vehicle, when the center of mass of the target convex-hull obstacle is on the left side of the leading line of the vehicle, mark the right-hand traffic on the target convex-hull obstacle label, when the center of mass of the target convex hull obstacle is on the right side of the leading line of the vehicle, mark the left traffic label on the target convex hull obstacle.
- the obstacle 2 is located on the leading line c of the own vehicle, at this time, the unmanned vehicle 100 only needs to follow the obstacle 2, and label the obstacle 2 with a follow label.
- the obstacle 3 is located in the road, but not on the leading line c of the own vehicle, and affects the safety of the lane of the driverless vehicle 100. At this time, it is necessary to pass from the left or right of the obstacle 3 to avoid the obstacle 3.
- the relative position of the center of mass of the obstacle 3 and the guiding line c of the vehicle is detected. If the center of mass of the obstacle 3 is located on the right side of the guiding line c of the vehicle (as shown in FIG.
- the second grid obstacle is converted into a second convex hull obstacle, that is, the grid Grid-type obstacles are converted into convex-hull-type obstacles to achieve unified decision-making for both grid-type and convex-hull type obstacles (that is, mixed-type obstacles), thereby simplifying obstacle decision-making for mixed-type obstacles
- the process accelerates the obstacle decision-making process, so that the decision-making planning module can conveniently and quickly make obstacle decisions.
- FIG. 8 is a partial flowchart of another vehicle decision planning method provided by an embodiment of the present disclosure.
- the method is applicable to the situation where the unmanned vehicle generates a driving area for static obstacles and/or dynamic obstacles, and the method can be executed by a driving space generator.
- generating a drivable area (or vehicle drivable area generation method) according to obstacle decision-making includes the following steps:
- the environment perception information includes at least two of lane information, obstacle information and vehicle information
- the obstacle information includes static obstacle information and/or dynamic obstacle information.
- the lane information may include lane line information and road boundary information, which may be acquired by a vehicle-mounted camera
- the obstacle information may include obstacle position information, obstacle size information, and obstacle motion information, wherein the obstacle position Information can be obtained by using high-precision maps and vehicle-mounted cameras/lidars, obstacle size information can be obtained by vehicle-mounted cameras, obstacle movement information can be obtained by vehicle-mounted cameras and/or laser radars
- vehicle information can include vehicle location information and vehicle motion information, among which, the vehicle position information can be obtained by high-precision maps and vehicle positioning modules (such as GPS), and the vehicle motion information can be obtained by vehicle motion sensors (such as speed sensors and acceleration sensors, etc.) Obtain.
- S320 Determine lane decision semantic information for each lane based on the environment perception information.
- the semantic information of lane decision includes passing time cost and safety cost.
- the passing time cost is used to characterize the traffic situation of the lane, for example, if the vehicle can pass through a lane quickly, the passing time of the lane is fast; the safety cost is used to characterize the safety of the lane.
- the passing time cost of each lane can be determined according to the magnitude relationship between the vehicle's longitudinal speed and the obstacle's longitudinal speed.
- the lane decision semantic information includes the passing time cost
- the lane decision semantic information of each lane is determined based on the environment perception information, including: for each lane, based on the environment perception information, determining The collision time of an obstacle; the collision time is determined as the passing time cost.
- the environmental perception information includes the position information of the vehicle and the longitudinal velocity information of the vehicle, as well as the obstacle position information and the longitudinal velocity information of the obstacle closest to the vehicle in each lane.
- the position information of the vehicle and the obstacle According to the longitudinal speed information of the vehicle and the longitudinal speed information of the obstacle, it is judged whether the longitudinal speed of the obstacle is less than the longitudinal speed of the vehicle.
- the longitudinal speed of the obstacle is less than the longitudinal speed of the vehicle, according to the longitudinal distance, the longitudinal speed information of the vehicle and the longitudinal speed information of the obstacle, the collision time when the vehicle collides with the obstacle in front is predicted, and the collision time is determined as the passing time cost.
- the preset time length is determined as the passing time cost. Based on the above technical solution, the following formula can be used to calculate the passing time cost:
- TCC is the passing time cost
- v adv is the longitudinal velocity of the vehicle
- v obs is the longitudinal velocity of the obstacle
- tcc max is the preset duration, which is a fixed value and is greater than the collision time, for example, 1000 (here is only a numerical value , the unit is the same as the collision time, such as seconds or milliseconds, etc.).
- the first obstacle in front of the vehicle is obstacle 1, and in the adjacent lane, the first obstacle in front of the vehicle is obstacle 2.
- the longitudinal velocity of the vehicle is 5m/s
- the longitudinal velocity of obstacle 1 is 1m/s
- the longitudinal velocity of obstacle 2 is 10m/s.
- the longitudinal velocity of the obstacle 1 is less than the longitudinal velocity of the vehicle, and the vehicle will collide with the obstacle 1.
- the distance D between the vehicle and the obstacle 1 is determined to be 16m, and it can be determined according to the above formula
- the collision time between the vehicle and the obstacle 1 is 4s, therefore, the passing time cost of the vehicle lane is 4.
- the passing time cost of the adjacent lane is the preset time length, such as 10000. Therefore, it can be determined that the passing time cost of the adjacent lane is greater than the passing time cost of the own vehicle lane, that is, the passability of the adjacent lane is better.
- the safety cost of each lane needs to be determined at the same time.
- the lane decision semantic information includes the safety cost
- the lane decision semantic information of each lane is determined based on the environment perception information, including: based on the lane information and the vehicle information, determining the vehicle lane and other lanes; for the vehicle For the lane, the first preset safety cost is determined as the safety cost; for other lanes, if it is determined based on the environmental perception information that the obstacle enters the danger zone of the vehicle within the preset time, the second preset safety cost is determined For the safety cost, if it is determined based on the environmental perception information that the obstacle does not enter the danger zone of the vehicle within the preset time, the first preset safety cost is determined as the safety cost, wherein the second preset safety cost It is different from the first preset security cost.
- the environmental perception information includes lane information, own vehicle information and obstacle information. Based on the lane information and own vehicle information, the own vehicle lane and other lanes are determined. For the own vehicle lane, the own vehicle has absolute right of way by default, that is, the own vehicle Driveways are the safest. For other lanes, the safety judgment can be made on the obstacles in the observation area of the vehicle.
- the preset time When it is predicted that the obstacles in the observation area of the vehicle will enter the dangerous area of the vehicle within a certain period of time in the future (that is, the preset time), it means The safety of the lane where the obstacle is at the current moment is low; when it is predicted that the obstacle in the observation area of the vehicle will not enter the danger zone of the vehicle for a period of time in the future, it means that the lane where the obstacle is at the current moment is relatively safe .
- the safety of the lane corresponding to the second preset safety cost is lower than the safety of the lane corresponding to the first preset safety cost.
- the second preset security cost is smaller than the first preset security cost.
- a penalty mechanism may be used to assign values to the first preset security cost and the second preset security cost, for example, the first preset security cost is 0, and the second preset security cost is -100000.
- the ST map (longitudinal displacement-time map) can be used to determine whether obstacles in other lanes will enter the danger zone of the vehicle in the future.
- determine the ST map curve of the vehicle and the ST map curve of the obstacle based on the environmental perception information, determine the ST map curve of the vehicle and the ST map curve of the obstacle; determine the dangerous area of the vehicle based on the ST map curve of the vehicle; There is overlap in the dangerous area of the vehicle; if the ST map curve of the obstacle overlaps with the dangerous area of the vehicle within the preset time, it is determined that the obstacle enters the dangerous area of the vehicle within the preset time; otherwise, it is determined that the obstacle enters the dangerous area of the vehicle within the preset time did not enter the danger zone of the vehicle.
- the ST graph curve of the own vehicle is the curve indicated by the own vehicle in the figure
- the ST graph curve of the obstacle includes the obstacles 1, 2, 3 and 4 respectively indicated in the figure curve.
- the dangerous area of the vehicle includes the dangerous area behind the vehicle (the area corresponding to the section L2) and the dangerous area in front of the vehicle (the area corresponding to the section L3)
- the observation area of the vehicle includes the observation area behind the vehicle (the area corresponding to the section L1).
- the preset time is T_e.
- L1 is 100 meters
- L2 is 20 meters
- L3 is 10 meters
- L4 is 100 meters
- T_e is 6 seconds. Referring to Fig.
- the ST diagram curve of the obstacle corresponding to obstacle 2 in the observation area behind the vehicle overlaps with the dangerous area behind the vehicle within the preset time T_e
- the obstacle ST diagram curve corresponding to obstacle 1 in the observation area behind the vehicle does not overlap with the dangerous area behind the vehicle within the preset time T_e
- the obstacle ST diagram curve corresponding to obstacle 3 in the observation area in front of the vehicle There is an overlap with the dangerous area ahead of the vehicle within the preset time T_e
- the obstacle ST map curve corresponding to the obstacle 4 in the observation area ahead of the vehicle does not overlap with the dangerous area ahead of the vehicle within the preset time T_e.
- the safety cost of the corresponding lane is the first 2.
- Preset safety cost if obstacle 1 and obstacle 4 do not enter the danger zone of the vehicle within the preset time, the safety of the lane where obstacle 1 and obstacle 4 are located at the current moment is relatively high, that is, the safety of the corresponding lane
- the sex cost is the first preset security cost.
- the overlapping of the obstacle ST map curve with the own vehicle's danger zone includes the obstacle ST map curve being completely located in the own vehicle's danger zone, or a part of the obstacle ST map curve being located in the own vehicle's danger zone.
- the obstacle is set to move at a constant speed
- the ST graph curve of the obstacle is a straight line
- the dangerous area of the vehicle is a parallelogram, all of which are convex-hull graphics.
- the embodiments of the present disclosure can simultaneously generate a drivable area based on the passing time cost and the safety cost, thereby selecting a lane that takes both passing and safety into account.
- each cost in the lane decision semantic information may be weighted and summed; based on the weighted sum result, a drivable area is generated.
- a weighted sum is performed on the passing time cost and the security cost to obtain a weighted sum result. For example:
- f is the weighted sum result (or weighted sum value)
- f pass is the passing time cost
- f safe is the security cost
- w 1 is the weight of the passing time cost
- w 2 is the weight of the safety cost.
- w 1 and w 2 can be obtained according to simulation or actual vehicle test experiments. Based on this technical solution, the embodiment of the present disclosure may determine the lane with the largest weighted sum value as the drivable area.
- the boundary of the drivable area is discretized to form drivable area boundary points, including left boundary points and right boundary points.
- the drivable area may be discretized with a fixed resolution based on the Freyner coordinate system.
- the left and right boundaries of the drivable area are generated with a fixed resolution based on the lane decision information and the curved Freyner coordinate system constructed by the center of the lane.
- the left border represents L in the Freyner coordinate system
- the upper and right bounds of the values represent the lower bounds of the L values in the Freyner coordinate system.
- the vertical distance between two adjacent left boundary points or two adjacent right boundary points is the above-mentioned fixed resolution.
- Figure 11 shows that the vehicle is in the right lane, and the result of the lane decision is also the right lane, so the left and right boundaries of the drivable area are shown in Figure 11. If the result of the lane decision is to change lanes, then the drivable area at this time includes two lanes.
- the vehicle drivable area generation method determines the lane decision semantic information of each lane according to the environmental perception information, and converts the lane decision semantic information into the constraint boundary of the drivable area, taking into account passability and safety, and can quickly Generate a drivable area with high passability and safety, accelerate the generation of driving trajectories, and realize fast avoidance of obstacles; at the same time, both the passing time cost and the safety cost can represent the passing cost of dynamic obstacles, and the passing time cost It can also characterize the passage cost of static obstacles. Therefore, the technical solution of the present disclosure generates a drivable area based on the passage time cost and safety cost, and can realize the passage planning of dynamic obstacles and static obstacles at the same time, and is applicable to dynamic environments Handling of obstacles.
- the lane decision semantic information also includes a traffic width cost
- determining the lane decision semantic information of each lane based on the environment perception information includes: determining the minimum traffic width of the lane based on the lane information and static obstacle information; The minimum traffic width is determined as the traffic width cost.
- the traffic width cost is used to represent the congestion of the lane by the static obstacles in front of the vehicle.
- determining the minimum passage width of the lane based on the lane information and the static obstacle information includes: determining the maximum passage width of each static obstacle on the lane based on the lane information and the static obstacle information; The minimum value of the maximum traffic width of is determined as the minimum traffic width of the lane.
- each static obstacle is projected into the Freyner coordinate system, and an SL bounding box of each obstacle is generated.
- calculate the left and right passage widths of each static obstacle determine the maximum passage width of each static obstacle from the left and right passage widths, and determine the maximum passage width from all static obstacles
- the greater the cost of the traffic width the smaller the degree of blockage of the lane by static obstacles. Exemplarily, as shown in FIG.
- the vehicle lane includes obstacle 1 and obstacle 2
- the adjacent lane includes obstacle 3 (obstacle 1, obstacle 2, and obstacle 3 are all static obstacles)
- the maximum passage width of obstacle 1 is d1
- the maximum passage width of obstacle 2 is d2
- the maximum passage width of obstacle 3 is d3, where d1 is smaller than d2 and d2 is smaller than d3, therefore, the minimum passage width of the vehicle lane is d1, that is, the traffic width cost of the vehicle lane is d1, and the minimum traffic width of the adjacent lane is d3, that is, the traffic width cost of the adjacent lane is d3.
- the degree of obstacle congestion in the adjacent lane is smaller than the degree of obstacle congestion in the own vehicle lane. Therefore, adjacent lanes can be further selected to generate drivable areas.
- the optimal drivable area cannot be determined based on the costs in the lane decision semantic information, in order to ensure the stability of the vehicle trajectory, the optimal selection of this area is made by adding the stability cost to the lane decision semantic information.
- Vehicle lanes generate drivable areas.
- the lane decision semantic information also includes a stability cost, and based on the environment perception information, determining the lane decision semantic information of each lane includes: based on the lane information and the vehicle information, determining the vehicle lane and other lane; for the vehicle lane, the first preset stability cost is determined as the stability cost; for other lanes, the second preset stability cost is determined as the stability cost, wherein the second preset stability cost is the same as the first A default stability cost is different.
- the first preset stability cost may be greater than the second preset stability cost, wherein the first preset stability cost may be 100, and the second preset stability cost may be 0.
- each cost in the lane decision semantic information is weighted and summed to obtain the weighted sum result, which can be calculated by the following formula:
- f narrow is the traffic width cost
- f stable is the stability cost
- w 3 is the weight of the traffic width cost
- w 4 is the weight of the stability cost
- the method further includes at least one of the following:
- the obstacle decision semantic information includes obstacles passing on the left side or obstacles passing on the right side.
- the preset traffic rules may include common traffic rules such as dashed and solid yellow lines, dashed and solid white lines, and lane indicator lines.
- the vehicle will choose to change lanes, so the two-lane lane during the lane change process is firstly used as the drivable area, that is, the original drivable area.
- the drivable area will be trimmed based on the preset traffic rules, and the drivable area defined by the black circles in Figure 13 is obtained, that is, the updated drivable area.
- the technical solution of the present disclosure may also update the drivable area based on the kinematics and dynamics constraints of the vehicle.
- the drivable area is updated by adding an additional drivable area when the vehicle temporarily takes the road.
- the included angle between the heading angle of the vehicle and the heading angle of the road network is ⁇
- the curvature of the vehicle’s lane position point is k r
- k ADV is the trajectory curvature of the vehicle calculated by using the wheel angle ⁇ of the unmanned bicycle model, and the wheelbase of the vehicle is B, then k′ r is the curvature change rate of the road network.
- k′ r 0.
- the final Freyner lateral velocity of the vehicle is 0, so based on the Freyner coordinates Department of Kinesiology has In this way, the drivable area is expanded outwards based on the calculated additional drivable area, so as to update the drivable area.
- the absolute safety of the drivable area cannot be guaranteed. Therefore, the drivable area may be affected by dynamic obstacles. The area is trimmed to ensure the safety of the remaining drivable area.
- the obstacle semantic information may include information used to characterize the movement state of the obstacle, such as obstacle merging, obstacle crossing, obstacle parallel driving, and reverse driving. Based on the semantic information of obstacles, the car will automatically judge whether to avoid obstacles laterally.
- the vehicle does not need to avoid the obstacle laterally; if it is determined based on the obstacle semantic information that the obstacle is too close to the vehicle lane, the vehicle needs to avoid the obstacle laterally.
- a preset safety area is added on both sides of the drivable area (if the drivable area is adjacent to the road boundary, the preset safety area can only be increased inside the drivable area), and the obstacle prediction module is judged (The previous module, not involved in this disclosure) Whether the output trajectory of the obstacle occupies the preset safety zone, if the movement trajectory occupies the preset security zone, trim the part of the drivable area that occupies the corresponding position of the preset security zone , as shown in the clipping area in Figure 15, to update the drivable area.
- the disclosed technical solution combines obstacle decision semantic information and ray tracing algorithms to accurately determine where static obstacles are located. area.
- update the drivable area based on static obstacle information, obstacle decision semantic information and ray tracing algorithm including: determine the collision point between light and obstacles based on static obstacle information, obstacle decision semantic information and ray tracing algorithm , the collision point is within the drivable area; based on the collision point, the drivable area is updated.
- determining the collision point between the light and the obstacle includes: determining the light source point and the ray projection direction based on the obstacle decision semantic information; Obstacle information, determine the range of light projection; based on the light source point, light projection direction and light projection range, scan the obstacle with light; determine the collision point between the light and the obstacle.
- the ray tracing algorithm may use a ray casting algorithm based on the Gilbert–Johnson–Keerthi algorithm to improve solution accuracy.
- the obstacle decision semantic information includes passing from the left side of the obstacle or passing from the right side of the obstacle.
- the obstacle decision semantic information is passing from the left side of the obstacle, it is determined that the light source point is located on the left side of the static obstacle, and the light projection direction is perpendicular to the direction of the lane and towards the static obstacle; when the obstacle decision semantic information is from the obstacle
- the object passes on the right side it is determined that the light source point is located on the right side of the static obstacle, and the light projection direction is perpendicular to the direction of the lane and facing the static obstacle.
- the static obstacle information such as the location information and size information of the static obstacle, the area where the static obstacle is located can be determined, thereby determining the ray projection range.
- the drivable area defined by each collision point is trimmed off.
- the embodiment of the present disclosure does not limit the specific position of the light source point, and in some embodiments, the light source point may be located on a border point of the drivable area.
- the embodiment of the present disclosure only uses the frenet bounding box to determine the ray projection range that can include the entire static obstacle, so as to ensure that the ray scans the obstacle completely, and the subsequently determined collision point is located on the static obstacle instead of Located on the boundary of the frenet bounding box.
- there are two static obstacles in the drivable area namely obstacle 1 and obstacle 2.
- obstacle decision semantic information of obstacle 1 it can be determined that the vehicle passes on the right side of obstacle 1
- obstacle decision semantic information of obstacle 2 it can be determined that the vehicle passes on the left side of obstacle 2.
- obstacle 1 based on the obstacle decision semantic information of obstacle 1, determine that the point light source is located at the right boundary point of the drivable area, and based on the static obstacle information of obstacle 1, determine the light projection range of the light source on obstacle 1 , that is, the above-mentioned ID range, so that the point light source can scan the obstacle 1 in sequence according to the ID range. Specifically, when the light collides with the obstacle 1, if the collision point is within the drivable area, the drivable area on the side away from the point light source is trimmed until the scan of the ray projection range is completed.
- the solution accuracy of the area occupied by static obstacles in the drivable area can be improved, and the trafficability of the vehicle can be improved.
- the embodiment of the present disclosure also provides a vehicle decision planning device.
- the vehicle decision planning device includes a base coordinate system generator 111 , a guiding line generator 112 , an obstacle decision maker 113 and a driving space generator 114 .
- the base coordinate system generator 111 is used to generate the base coordinate system;
- the guide line generator 112 is used to generate the guide line under the base coordinate system to determine a general driving trajectory of the vehicle in the future;
- the obstacle decision maker 113 used to make obstacle decisions under the constraints of the guide line;
- the driving space generator 114 used to generate a drivable area according to the obstacle decisions.
- Fig. 18 is a block diagram of functional modules of an obstacle decision maker in a vehicle decision planning device provided by an embodiment of the present disclosure.
- the obstacle decision maker (or the decision-making device for avoiding obstacles) includes an information acquisition module 401 , a preprocessing module 402 , a type conversion module 403 and an avoidance decision-making module 404 .
- the information acquiring module 401 is configured to acquire road information, first grid obstacle information of the first grid obstacle, and first convex hull obstacle information of the first convex hull obstacle;
- the preprocessing module 402 is configured to preprocess the obstacles in the first grid based on the road information and the obstacle information in the first grid to obtain the obstacles in the second grid, wherein the number of obstacles in the second grid is less than the number of obstacles in the first grid The number of grid obstacles;
- a type conversion module 403, configured to convert the second grid obstacle into a second convex hull obstacle
- An avoidance decision-making module 404 configured to make an avoidance decision for the target convex-hull obstacle based on the target convex-hull obstacle information of the target convex-hull obstacle, wherein the target convex-hull obstacle includes the first convex-hull obstacle and/or the second convex-hull obstacle Two convex hull obstacles.
- the preprocessing module 402 includes:
- an obstacle filtering unit configured to filter out first grid obstacles located outside the road based on the road information and the first grid obstacle information; use the remaining first grid obstacles as second grid obstacles; and /or,
- the obstacle aggregation unit is configured to determine the first grid obstacle located in the road based on the road information and the first grid obstacle information; perform aggregation processing on the first grid obstacle located in the road; The first grid obstacle acts as the second grid obstacle.
- the obstacle filtering unit includes:
- the road bounding box generation subunit is used to generate a road bounding box along the road traffic direction based on road information
- a grid obstacle bounding box generating subunit configured to generate a grid obstacle bounding box of the first grid obstacle based on the information of the first grid obstacle;
- the first grid obstacle subunit is used to determine the first grid obstacle located outside the road based on the grid obstacle bounding box and the road bounding box;
- the first grid obstacle filtering subunit is configured to filter out first grid obstacles outside the road.
- the road bounding box generating subunit is specifically used for:
- the road boundary is discretized into boundary points
- a road bounding box is generated.
- the road bounding box generating subunit is specifically used for:
- the road boundary is discretized, and the boundary point groups arranged at intervals along the road traffic direction are obtained.
- Each group of boundary point groups includes the corresponding left boundary point and right boundary point in the horizontal direction, and the horizontal direction is perpendicular to the road direction of travel.
- the road bounding box generating subunit is specifically used for:
- a rectangular frame passing through each boundary point in any two adjacent boundary point groups is generated, and the rectangular frame is used as a road bounding box.
- the road bounding box generating subunit is specifically used for:
- the grid obstacle bounding box generating subunit is specifically used for:
- the first grid obstacle subunit is specifically used for:
- For each grid obstacle bounding box based on the grid obstacle bounding box and the road bounding box, determine the target road bounding box with the smallest Euclidean distance from the road bounding box to the grid obstacle bounding box;
- the grid obstacle bounding box does not collide with the corresponding target road bounding box, it is determined that the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.
- the device also includes:
- the first grid obstacle position judging module is configured to judge whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road if the grid obstacle bounding box collides with the corresponding target road bounding box.
- the first grid obstacle position judging module is specifically used for:
- the collision detection is performed by vector cross product, and it is judged whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.
- the first grid obstacle position judging module is specifically used for:
- the first grid obstacle position judging module is specifically used for:
- the obstacle aggregation unit includes:
- the first obstacle bounding box generating subunit is configured to generate a first obstacle bounding box of the first grid obstacle based on the first grid obstacle information of the first grid obstacle located in the road;
- the second obstacle bounding box generation subunit is used to merge two adjacent first obstacle bounding boxes when the Euclidean distance between two adjacent first obstacle bounding boxes is less than the width of the vehicle to generate The second obstacle bounding box;
- the execution unit which is used to use the second obstacle bounding box as the first obstacle bounding box, return to execute when the Euclidean distance between two adjacent first obstacle bounding boxes is smaller than the width of the vehicle, the two adjacent first obstacle bounding boxes
- the first obstacle bounding box is merged to generate the second obstacle bounding box until the Euclidean distance between the second obstacle bounding box and the adjacent first obstacle bounding box is greater than or equal to the width of the vehicle, or there is no The first obstacle bounding box adjacent to the second obstacle bounding box.
- avoidance decision module 404 includes:
- the first decision-making unit is configured to, based on the target convex-hull obstacle information, label the target convex-hull obstacle that meets the preset filter condition without avoidance label or lateral avoidance label; and/or,
- the second decision-making unit is configured to label the target convex-hull obstacle with an avoidance label based on the information of the target convex-hull obstacle and the guidance line of the vehicle, wherein the avoidance label includes a left pass label, a right pass label or a follow label.
- the preset filtering conditions include at least one of the following:
- the target convex hull obstacle is located outside the road;
- the movement state of the target convex hull obstacle meets the condition of no need for lateral avoidance
- the target convex obstacle is located on the leading line of the vehicle.
- the first decision-making unit is specifically used to:
- no lateral avoidance conditions include any of the following:
- the target convex obstacle crosses the road
- the target convex hull obstacle changes to the lane of the vehicle
- the longitudinal speed of the target convex obstacle is greater than the speed of the vehicle.
- the second decision-making unit is specifically used to:
- target convex hull obstacle If the target convex hull obstacle is located on the leading line of the vehicle, mark the following label on the target convex hull obstacle;
- the target convex-hull obstacle is not located on the leading line of the vehicle, when the center of mass of the target convex-hull obstacle is on the left side of the guiding line of the own vehicle, the right-hand traffic label is placed on the target convex-hull obstacle; when the target convex-hull obstacle When the center of mass of the object is on the right side of the vehicle's guiding line, mark the left passing label on the target convex hull obstacle.
- Fig. 19 is a block diagram of functional modules of a driving space generator in a vehicle decision planning device provided by an embodiment of the present disclosure.
- the driving space generator (or vehicle drivable area generation device) includes a perception information acquisition module 501 , a lane decision semantic information determination module 502 and a drivable area generation module 503 .
- the perception information acquisition module 501 is used to obtain environment perception information, wherein the environment perception information includes at least two of lane information, obstacle information and own vehicle information, and the obstacle information includes static obstacle information and/or dynamic obstacle information item information;
- Lane decision semantic information determination module 502 configured to determine the lane decision semantic information of each lane based on the environment perception information, wherein the lane decision semantic information includes passing time cost and safety cost;
- a drivable area generation module 503 is configured to generate a drivable area based on the lane decision semantic information.
- the lane decision semantic information determination module 502 is specifically used to:
- the lane decision semantic information determination module 502 is also used for:
- the preset duration is determined as the passing time cost.
- the lane decision semantic information determination module 502 is specifically configured to:
- the first preset safety cost is determined as the safety cost
- the second preset safety cost is determined as the safety cost; Assuming that the vehicle has not entered the danger zone within a certain period of time, the first preset safety cost is determined as the safety cost, wherein the second preset safety cost is different from the first preset safety cost.
- the lane decision semantic information determination module 502 is specifically used to:
- the ST map curve of the obstacle overlaps with the dangerous zone of the vehicle within the preset time, it is determined that the obstacle has entered the dangerous zone of the vehicle within the preset time; otherwise, it is determined that the obstacle has not entered the dangerous zone of the vehicle within the preset time .
- the lane decision semantic information also includes a traffic width cost, and the lane decision semantic information determination module 502 is specifically used for:
- the lane decision semantic information determination module 502 is specifically used to:
- the minimum value among the maximum passage widths of the static obstacles is determined as the minimum passage width of the lane.
- the lane decision semantic information also includes a stability cost, and the lane decision semantic information determination module 502 is specifically used for:
- the first preset stability cost is determined as the stability cost
- a second preset stability cost is determined as the stability cost, wherein the second preset stability cost is different from the first preset stability cost.
- the drivable area generating module 503 is specifically used to:
- the above-mentioned device also includes:
- the discretization module is used to discretize the boundary of the drivable area after the drivable area is generated based on the lane decision semantic information.
- the above device further includes a drivable area update module, which is specifically used for at least one of the following update operations after generating the drivable area based on the lane decision semantic information:
- the obstacle decision semantic information includes passing from the left side of the obstacle or from the right side of the obstacle.
- the drivable area update module is specifically used for:
- the drivable area update module is specifically used for:
- the drivable area is updated.
- the drivable area update module is specifically used for:
- the vehicle decision-making planning device disclosed in the above embodiments can execute the vehicle decision-making planning method disclosed in the above embodiments, and has the same or corresponding beneficial effects. In order to avoid repetition, details are not repeated here.
- An embodiment of the present disclosure also provides an electronic device, including: a memory and one or more processors; wherein, the memory is connected to the one or more processors in communication, and the memory stores information that can be executed by the one or more processors. Instructions, when the instructions are executed by one or more processors, the electronic device is used to implement the method described in any embodiment of the present disclosure.
- FIG. 20 is a schematic structural diagram of an electronic device suitable for implementing the embodiments of the present disclosure.
- an electronic device 600 includes a central processing unit (CPU) 601, which can operate according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage section 608 into a random access memory (RAM) 603 Instead, various processes in the aforementioned embodiments are executed.
- the RAM 603 various programs and data necessary for the operation of the electronic device 600 are also stored.
- the CPU 601 , ROM 602 , and RAM 603 are connected to each other via a bus 604 .
- An input/output (I/O) interface 605 is also connected to the bus 604 .
- the following components are connected to the I/O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage section 608 including a hard disk, etc. and a communication section 609 including a network interface card such as a LAN card, a modem, or the like.
- the communication section 609 performs communication processing via a network such as the Internet.
- a drive 610 is also connected to the I/O interface 605 as needed.
- a removable medium 611 such as a magnetic disk, optical disk, magneto-optical disk, semiconductor memory, etc. is mounted on the drive 610 as necessary so that a computer program read therefrom is installed into the storage section 608 as necessary.
- the methods described above can be implemented as a computer software program.
- the embodiments of the present disclosure include a computer program product including a computer program tangibly embodied on its readable medium, the computer program including program codes for executing the aforementioned obstacle avoidance method.
- the computer program may be downloaded and installed from a network via the communication portion 609 and/or installed from a removable medium 611 .
- each block in a roadmap or block diagram may represent a module, program segment, or portion of code that contains one or more executable instruction.
- the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or they may sometimes be executed in the reverse order, depending upon the functionality involved.
- each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations can be implemented by a dedicated hardware-based system that performs the specified functions or operations , or may be implemented by a combination of dedicated hardware and computer instructions.
- the units or modules involved in the embodiments described in the present disclosure may be implemented by means of software or hardware.
- the described units or modules may also be set in the processor, and the names of these units or modules do not constitute limitations on the units or modules themselves in some cases.
- an embodiment of the present disclosure also provides a computer-readable storage medium, which may be the computer-readable storage medium contained in the device described in the above implementation manner; computer-readable storage media stored in the device.
- the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a computing device, they can be used to implement the method described in any embodiment of the present disclosure.
- a road bounding box along the road traffic direction including:
- the road bounding box is generated.
- the road boundary is discretized into boundary points, including:
- the road boundary is discretized to obtain boundary point groups arranged at intervals along the road traffic direction, wherein each group of boundary point groups includes corresponding left boundary points and right boundary points in the transverse direction, and the The transverse direction is perpendicular to the direction of road traffic.
- the road boundary is discretized to obtain boundary point groups arranged at intervals along the road traffic direction, including:
- generating a grid obstacle bounding box of the first grid obstacle includes:
- the grid obstacle bounding box is generated based on the grid obstacle outline.
- determining the first grid obstacle located outside the road includes:
- the grid obstacle bounding box does not collide with the corresponding target road bounding box, it is determined that the first grid obstacle corresponding to the grid obstacle bounding box is outside the road.
- the method also includes:
- the grid obstacle bounding box collides with the corresponding target road bounding box, it is determined whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.
- Judging whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road includes:
- the collision detection is performed by vector cross product, and it is judged whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road.
- Performing collision detection based on the boundary point of the target road bounding box and the grid obstacle bounding box by vector cross product, and judging whether the first grid obstacle corresponding to the grid obstacle bounding box is located outside the road include:
- the method also includes:
- Aggregating the first grid obstacles located in the road includes:
- the second obstacle bounding box as the first obstacle bounding box, return to execute when the Euclidean distance between two adjacent first obstacle bounding boxes is smaller than the vehicle width, set the two adjacent first obstacle bounding boxes
- the first obstacle bounding box is merged to generate the second obstacle bounding box until the Euclidean distance between the second obstacle bounding box and the adjacent first obstacle bounding box is greater than or equal to the width of the vehicle, or there is no The first obstacle bounding box adjacent to the second obstacle bounding box.
- an avoidance decision is made for the target convex-hull obstacle, including:
- the target convex-hull obstacle information Based on the target convex-hull obstacle information, label the target convex-hull obstacle that satisfies the preset filter condition without avoidance or lateral avoidance; and/or,
- an avoidance label is attached to the target convex obstacle, wherein the avoidance label includes a left passing label, a right passing label or a following label.
- the preset filtering conditions include at least one of the following:
- the target convex hull obstacle is located outside the road;
- the motion state of the target convex hull obstacle meets the condition of no need for lateral avoidance
- the target convex obstacle is located on the leading line of the vehicle.
- the target convex-hull obstacle that satisfies the preset filter condition is marked with no avoidance label or no lateral avoidance label, including:
- the conditions that do not require lateral avoidance include any of the following:
- the target convex hull obstacle crosses the road
- the target convex hull obstacle changes lanes to the own vehicle lane
- the longitudinal speed of the target convex hull obstacle is greater than the speed of the own vehicle.
- label the target convex hull obstacle with an avoidance label including:
- the follow-up label is applied to the target convex-hull obstacle
- the target convex obstacle is not located on the vehicle guide line, when the center of mass of the target convex obstacle is on the left side of the vehicle guide line, mark the target convex obstacle with As for the right-side traffic label, when the center of mass of the target convex-hull obstacle is on the right side of the vehicle guiding line, the left-side traffic label is attached to the target convex-hull obstacle.
- the method also includes:
- the preset time length is determined as the passing time cost.
- the method also includes:
- the ST map curve of the obstacle overlaps with the dangerous zone of the own vehicle within the preset time, it is determined that the obstacle enters the dangerous zone of the own vehicle within the preset time; otherwise, it is determined that the obstacle is within the dangerous zone of the own vehicle. Did not enter the danger zone of the vehicle within the preset time.
- the lane decision semantic information also includes traffic width cost, and based on the environment perception information, the lane decision semantic information of each lane is determined, including:
- the minimum passage width is determined as the passage width cost.
- determine the minimum traffic width of the lane including:
- the minimum value among the maximum passage widths of the static obstacles is determined as the minimum passage width of the lane.
- the lane decision semantic information also includes a stability cost, and based on the environment perception information, the lane decision semantic information of each lane is determined, including:
- a second preset stability cost is determined as the stability cost, wherein the second preset stability cost is different from the first preset stability cost.
- a drivable area is generated, including:
- the method further includes:
- the boundary of the drivable area is discretized.
- the method further includes at least one of the following:
- the drivable area is updated based on the static obstacle information, obstacle decision semantic information and a ray tracing algorithm, where the obstacle decision semantic information includes passing from the left side of the obstacle or passing from the right side of the obstacle.
- the drivable area is updated, including:
- the obstacle decision semantic information and the ray tracing algorithm determine the collision point between the light and the obstacle, and the collision point is located in the drivable area;
- the drivable area is updated.
- obstacle decision semantic information Based on the static obstacle information, obstacle decision semantic information and ray tracing algorithm, determine the collision point between the light and the obstacle, including:
- a decision-making method for avoiding obstacles comprising:
- the target convex-hull obstacle includes the first convex-hull obstacle and/or the The second convex hull obstacle.
- a method for generating a vehicle drivable area comprising:
- the environment awareness information includes at least two of lane information, obstacle information and vehicle information, and the obstacle information includes static obstacle information and/or dynamic obstacle information;
- lane decision semantic information Based on the environment perception information, determine lane decision semantic information of each lane, wherein the lane decision semantic information includes passing time cost and safety cost;
- a drivable area is generated.
- the disclosure converts grid-type obstacles into convex-hull-type obstacles, and realizes unified decision-making for grid-type and convex-hull-type obstacles, thereby simplifying the obstacle decision-making process for mixed-type obstacles and speeding up
- the obstacle decision-making process enables the decision-making planning module to make obstacle decisions conveniently and quickly, and has strong industrial applicability.
Landscapes
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Traffic Control Systems (AREA)
Abstract
Description
Claims (19)
- 一种车辆决策规划方法,其特征在于,包括:生成基坐标系;在所述基坐标系下生成引导线,以决策车辆未来的一个大致行驶轨迹;在所述引导线的约束下,进行障碍物决策;根据障碍物决策生成可行驶区域。
- 根据权利要求1所述的方法,其特征在于,进行障碍物决策,包括:获取道路信息、第一栅格障碍物的第一栅格障碍物信息和第一凸包障碍物的第一凸包障碍物信息;基于所述道路信息和所述第一栅格障碍物信息,对所述第一栅格障碍物进行预处理,得到第二栅格障碍物;将所述第二栅格障碍物转换成第二凸包障碍物;基于目标凸包障碍物的目标凸包障碍物信息,对所述目标凸包障碍物做出避让决策,其中,所述目标凸包障碍物包括所述第一凸包障碍物和/或所述第二凸包障碍物。
- 根据权利要求2所述的方法,其特征在于,基于所述道路信息和所述第一栅格障碍物信息,对所述第一栅格障碍物进行预处理,得到第二栅格障碍物,包括:基于所述道路信息和所述第一栅格障碍物信息,过滤掉位于道路之外的所述第一栅格障碍物;将剩余所述第一栅格障碍物作为所述第二栅格障碍物;和/或,基于所述道路信息和所述第一栅格障碍物信息,确定位于所述道路内的所述第一栅格障碍物;对位于所述道路内的所述第一栅格障碍物进行聚合处理;将聚合处理后的所述第一栅格障碍物作为所述第二栅格障碍物。
- 根据权利要求3所述的方法,其特征在于,基于所述道路信息和所述第一栅格障碍物信息,过滤掉位于道路之外的所述第一栅格障碍物,包括:基于所述道路信息,生成沿道路通行方向的道路包围盒;基于所述第一栅格障碍物信息,生成所述第一栅格障碍物的栅格障碍物包围盒;基于所述栅格障碍物包围盒和所述道路包围盒,确定位于道路之外的所述第一栅格障碍物;过滤掉位于道路之外的所述第一栅格障碍物。
- 根据权利要求1-4任一项所述的方法,其特征在于,根据障碍物决策生成可行驶区域,包括:获取环境感知信息,其中,所述环境感知信息包括车道信息、障碍物信息和本车信息中的至少两个,所述障碍物信息包括静态障碍物信息和/或动态障碍物信息;基于所述环境感知信息,确定每个车道的车道决策语义信息,其中,所述车道决策语义信息包括通过时间代价和安全性代价;基于所述车道决策语义信息,生成可行驶区域。
- 根据权利要求5所述的方法,其特征在于,当所述车道决策语义信息包括通过时间代价时,基于所述环境感知信息,确定每个车道的车道决策语义信息,包括:对于每个车道,基于所述环境感知信息,确定本车与本车前方第一个障碍物的碰撞时间;将所述碰撞时间确定为所述通过时间代价。
- 根据权利要求5所述的方法,其特征在于,当所述车道决策语义信息包括安全性代价时,基于所述环境感知信息,确定每个车道的车道决策语义信息,包括:基于所述车道信息和所述本车信息,确定本车车道和其他车道;对于所述本车车道,将第一预设安全性代价确定为所述安全性代价;对于所述其他车道,如果基于所述环境感知信息,确定障碍物在预设时间内进入本车危险区,则将第二预设安全性代价确定为所述安全性代价,如果基于所述环境感知信息,确定所述障碍物在所述预设时间内未进入本车危险区,则将所述第一预设安全性代价确定为所述安全性代价,其中,所述第二预设安全性代价与所述第一预设安全性代价不同。
- 一种避让障碍物的决策方法,其特征在于,包括:获取道路信息、第一栅格障碍物的第一栅格障碍物信息和第一凸包障碍物的第一凸包障碍物信息;基于所述道路信息和所述第一栅格障碍物信息,对所述第一栅格障碍物进行预处理,得到第二栅格障碍物;将所述第二栅格障碍物转换成第二凸包障碍物;基于目标凸包障碍物的目标凸包障碍物信息,对所述目标凸包障碍物做出避让决策,其中,所述目标凸包障碍物包括所述第一凸包障碍物和/或所述第二凸包障碍物。
- 根据权利要求8所述的方法,其特征在于,基于所述道路信息和所述第一栅格障碍物信息,对所述第一栅格障碍物进行预处理,得到第二栅格障碍物,包括:基于所述道路信息和所述第一栅格障碍物信息,过滤掉位于道路之外的所述第一栅格障碍物;将剩余所述第一栅格障碍物作为所述第二栅格障碍物;和/或,基于所述道路信息和所述第一栅格障碍物信息,确定位于所述道路内的所述第一栅格障碍物;对位于所述道路内的所述第一栅格障碍物进行聚合处理;将聚合处理后的所述第一栅格障碍物作为所述第二栅格障碍物。
- 一种车辆可行驶区域生成方法,其特征在于,包括:获取环境感知信息,其中,所述环境感知信息包括车道信息、障碍物信息和本车信息中的至少两个,所述障碍物信息包括静态障碍物信息和/或动态障碍物信息;基于所述环境感知信息,确定每个车道的车道决策语义信息,其中,所述车道决策语义信息包括通过时间代价和安全性代价;基于所述车道决策语义信息,生成可行驶区域。
- 根据权利要求10所述的方法,其特征在于,当所述车道决策语义信息包括通过时间代价时,基于所述环境感知信息,确定每个车道的车道决策语义信息,包括:对于每个车道,基于所述环境感知信息,确定本车与本车前方第一个障碍物的碰撞时间;将所述碰撞时间确定为所述通过时间代价。
- 根据权利要求10所述的方法,其特征在于,当所述车道决策语义信息包括安全性代价时,基于所述环境感知信息,确定每个车道的车道决策语义信息,包括:基于所述车道信息和所述本车信息,确定本车车道和其他车道;对于所述本车车道,将第一预设安全性代价确定为所述安全性代价;对于所述其他车道,如果基于所述环境感知信息,确定障碍物在预设时间内进入本车危险区,则将第二预设安全性代价确定为所述安全性代价,如果基于所述环境感知信息,确定所述障碍物在所述预设时间内未进入本车危险区,则将所述第一预设安全性代价确定为所述安全性代价,其中,所述第二预设安全性代价与所述第一预设安全性代价不同。
- 一种车辆决策规划装置,其特征在于,包括:基坐标系生成器,用于生成基坐标系;引导线生成器,用于在所述基坐标系下生成引导线,以决策车辆未来的一个大致行驶轨迹;障碍物决策器,用于在所述引导线的约束下,进行障碍物决策;行驶空间生成器,用于根据障碍物决策生成可行驶区域。
- 根据权利要求13所述的装置,其特征在于,障碍物决策器包括:信息获取模块,用于获取道路信息、第一栅格障碍物的第一栅格障碍物信息和第一凸包障碍物的第一凸包障碍物信息;预处理模块,用于基于所述道路信息和所述第一栅格障碍物信息,对所述第一栅格障碍物进行预处理,得到第二栅格障碍物,其中,所述第二栅格障碍物的数量小于所述第一栅格障碍物的数量;类型转换模块,用于将所述第二栅格障碍物转换成第二凸包障碍物;避让决策模块,用于基于目标凸包障碍物的目标凸包障碍物信息,对所述目标凸包障碍物做出避让决策,其中,所述目标凸包障碍物包括所述第一凸包障碍物和/或所述第二凸包障碍物。
- 根据权利要求13所述的装置,其特征在于,行驶空间生成器包括:感知信息获取模块,用于获取环境感知信息,其中,所述环境感知信息包括车道信息、障碍物信息和本车信息中的至少两个,所述障碍物信息包括静态障碍物信息和/或动态障碍物信息;车道决策语义信息确定模块,用于基于所述环境感知信息,确定每个车道的车道决策语义信息,其中,所述车道决策语义信息包括通过时间代价和安全性代价;可行驶区域生成模块,用于基于所述车道决策语义信息,生成可行驶区域。
- 一种避让障碍物的决策装置,其特征在于,包括:信息获取模块,用于获取道路信息、第一栅格障碍物的第一栅格障碍物信息和第一凸包障碍物的第一凸包障碍物信息;预处理模块,用于基于所述道路信息和所述第一栅格障碍物信息,对所述第一栅格障碍物进行预处理,得到第二栅格障碍物,其中,所述第二栅格障碍物的数量小于所述第一栅格障碍物的数量;类型转换模块,用于将所述第二栅格障碍物转换成第二凸包障碍物;避让决策模块,用于基于目标凸包障碍物的目标凸包障碍物信息,对所述目标凸包障碍物做出避让决策,其中,所述目标凸包障碍物包括所述第一凸包障碍物和/或所述第二凸包障碍物。
- 一种车辆可行驶区域生成装置,其特征在于,包括:感知信息获取模块,用于获取环境感知信息,其中,所述环境感知信息包括车道信息、障碍物信息和本车信息中的至少两个,所述障碍物信息包括静态障碍物信息和/或动态障碍物信息;车道决策语义信息确定模块,用于基于所述环境感知信息,确定每个车道的车道决策语义信息,其中,所述车道决策语义信息包括通过时间代价和安全性代价;可行驶区域生成模块,用于基于所述车道决策语义信息,生成可行驶区域。
- 一种电子设备,其特征在于,包括:存储器以及一个或多个处理器;其中,所述存储器与所述一个或多个处理器通信连接,所述存储器中存储有可被所述一个或多个处理器执行的指令,所述指令被所述一个或多个处理器执行时,所述电子设备用于实现如权利要求1-12中任一项所述的方法。
- 一种计算机可读存储介质,其上存储有计算机可执行指令,其特征在于,当所述计算机可执行指令被计算装置执行时,可用来实现如权利要求1-12中任一项所述的方法。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP22859896.7A EP4368465A1 (en) | 2021-08-25 | 2022-04-22 | Vehicle decision-making planning method and apparatus, and device and medium |
KR1020247002509A KR20240025632A (ko) | 2021-08-25 | 2022-04-22 | 차량 의사결정 계획 방법, 장치, 기기 및 매체 |
US18/433,445 US20240174221A1 (en) | 2021-04-25 | 2024-02-06 | Vehicle decision-making planning method and apparatus, device and medium |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110984268.3A CN113682300B (zh) | 2021-08-25 | 2021-08-25 | 避让障碍物的决策方法、装置、设备及介质 |
CN202110984268.3 | 2021-08-25 | ||
CN202111095293.2A CN115817464A (zh) | 2021-09-17 | 2021-09-17 | 车辆可行驶区域生成方法、装置、设备及介质 |
CN202111095293.2 | 2021-09-17 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US18/433,445 Continuation US20240174221A1 (en) | 2021-04-25 | 2024-02-06 | Vehicle decision-making planning method and apparatus, device and medium |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2023024542A1 true WO2023024542A1 (zh) | 2023-03-02 |
Family
ID=85322344
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2022/088532 WO2023024542A1 (zh) | 2021-04-25 | 2022-04-22 | 车辆决策规划方法、装置、设备及介质 |
Country Status (4)
Country | Link |
---|---|
US (1) | US20240174221A1 (zh) |
EP (1) | EP4368465A1 (zh) |
KR (1) | KR20240025632A (zh) |
WO (1) | WO2023024542A1 (zh) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116502479A (zh) * | 2023-06-29 | 2023-07-28 | 之江实验室 | 一种三维物体在仿真环境中的碰撞检测方法和装置 |
CN117068199A (zh) * | 2023-08-08 | 2023-11-17 | 广州汽车集团股份有限公司 | 车辆可行驶空间的生成方法、装置、车辆及存储介质 |
CN117341683A (zh) * | 2023-12-04 | 2024-01-05 | 苏州观瑞汽车技术有限公司 | 一种基于多目标识别的车辆动态轨迹拟合避障方法及*** |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110550029A (zh) * | 2019-08-12 | 2019-12-10 | 华为技术有限公司 | 障碍物避让方法及装置 |
CN110562258A (zh) * | 2019-09-30 | 2019-12-13 | 驭势科技(北京)有限公司 | 一种车辆自动换道决策的方法、车载设备和存储介质 |
US20200089239A1 (en) * | 2018-09-14 | 2020-03-19 | The Boeing Company | Computer-Implemented Method and a System for Defining a Path for a Vehicle Within an Environment With Obstacles |
CN111780777A (zh) * | 2020-07-13 | 2020-10-16 | 江苏中科智能制造研究院有限公司 | 一种基于改进a*算法和深度强化学习的无人车路径规划方法 |
CN112068545A (zh) * | 2020-07-23 | 2020-12-11 | 哈尔滨工业大学(深圳) | 一种无人驾驶车辆在十字路口的行驶轨迹规划方法、***及存储介质 |
CN112319477A (zh) * | 2020-11-02 | 2021-02-05 | 天津大学 | 一种用于无人驾驶的决策规划方法 |
CN113682300A (zh) * | 2021-08-25 | 2021-11-23 | 驭势科技(北京)有限公司 | 避让障碍物的决策方法、装置、设备及介质 |
-
2022
- 2022-04-22 WO PCT/CN2022/088532 patent/WO2023024542A1/zh active Application Filing
- 2022-04-22 EP EP22859896.7A patent/EP4368465A1/en active Pending
- 2022-04-22 KR KR1020247002509A patent/KR20240025632A/ko active Search and Examination
-
2024
- 2024-02-06 US US18/433,445 patent/US20240174221A1/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20200089239A1 (en) * | 2018-09-14 | 2020-03-19 | The Boeing Company | Computer-Implemented Method and a System for Defining a Path for a Vehicle Within an Environment With Obstacles |
CN110550029A (zh) * | 2019-08-12 | 2019-12-10 | 华为技术有限公司 | 障碍物避让方法及装置 |
CN110562258A (zh) * | 2019-09-30 | 2019-12-13 | 驭势科技(北京)有限公司 | 一种车辆自动换道决策的方法、车载设备和存储介质 |
CN111780777A (zh) * | 2020-07-13 | 2020-10-16 | 江苏中科智能制造研究院有限公司 | 一种基于改进a*算法和深度强化学习的无人车路径规划方法 |
CN112068545A (zh) * | 2020-07-23 | 2020-12-11 | 哈尔滨工业大学(深圳) | 一种无人驾驶车辆在十字路口的行驶轨迹规划方法、***及存储介质 |
CN112319477A (zh) * | 2020-11-02 | 2021-02-05 | 天津大学 | 一种用于无人驾驶的决策规划方法 |
CN113682300A (zh) * | 2021-08-25 | 2021-11-23 | 驭势科技(北京)有限公司 | 避让障碍物的决策方法、装置、设备及介质 |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116502479A (zh) * | 2023-06-29 | 2023-07-28 | 之江实验室 | 一种三维物体在仿真环境中的碰撞检测方法和装置 |
CN116502479B (zh) * | 2023-06-29 | 2023-09-01 | 之江实验室 | 一种三维物体在仿真环境中的碰撞检测方法和装置 |
CN117068199A (zh) * | 2023-08-08 | 2023-11-17 | 广州汽车集团股份有限公司 | 车辆可行驶空间的生成方法、装置、车辆及存储介质 |
CN117068199B (zh) * | 2023-08-08 | 2024-05-24 | 广州汽车集团股份有限公司 | 车辆可行驶空间的生成方法、装置、车辆及存储介质 |
CN117341683A (zh) * | 2023-12-04 | 2024-01-05 | 苏州观瑞汽车技术有限公司 | 一种基于多目标识别的车辆动态轨迹拟合避障方法及*** |
CN117341683B (zh) * | 2023-12-04 | 2024-04-23 | 苏州观瑞汽车技术有限公司 | 一种基于多目标识别的车辆动态轨迹拟合避障方法及*** |
Also Published As
Publication number | Publication date |
---|---|
KR20240025632A (ko) | 2024-02-27 |
US20240174221A1 (en) | 2024-05-30 |
EP4368465A1 (en) | 2024-05-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11970168B2 (en) | Vehicle trajectory modification for following | |
US11875681B2 (en) | Drive envelope determination | |
CN112068545B (zh) | 一种无人驾驶车辆在十字路口的行驶轨迹规划方法、***及存储介质 | |
US11532167B2 (en) | State machine for obstacle avoidance | |
US20240149868A1 (en) | Collision prediction and avoidance for vehicles | |
US11225247B2 (en) | Collision prediction and avoidance for vehicles | |
US11427191B2 (en) | Obstacle avoidance action | |
WO2023024542A1 (zh) | 车辆决策规划方法、装置、设备及介质 | |
CN111874006B (zh) | 路线规划处理方法和装置 | |
CN112789481A (zh) | 对自上而下场景的轨迹预测 | |
WO2020180881A1 (en) | State machine for traversing junctions | |
CN114258366A (zh) | 对于自主车辆的折线轮廓表示 | |
JP2017146730A (ja) | 経路決定装置 | |
EP4086875A1 (en) | Self-driving method and related device | |
KR20190045308A (ko) | 차량 판정 방법, 주행 경로 보정 방법, 차량 판정 장치, 및 주행 경로 보정 장치 | |
US11643105B2 (en) | Systems and methods for generating simulation scenario definitions for an autonomous vehicle system | |
EP4052174A1 (en) | Obstacle avoidance action | |
US20230132512A1 (en) | Autonomous vehicle trajectory determination based on state transition model | |
WO2021225822A1 (en) | Trajectory classification | |
US12017645B1 (en) | Controlling merging vehicles | |
JP7205804B2 (ja) | 車両制御装置 | |
WO2023047148A1 (ja) | 走行支援方法及び 走行支援装置 | |
CN115817464A (zh) | 车辆可行驶区域生成方法、装置、设备及介质 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22859896 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 20247002509 Country of ref document: KR Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1020247002509 Country of ref document: KR |
|
ENP | Entry into the national phase |
Ref document number: 2024505119 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2022859896 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref document number: 2022859896 Country of ref document: EP Effective date: 20240207 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |