CN116519316A - Automatic driving test method, device, storage medium and equipment based on behavior tree - Google Patents
Automatic driving test method, device, storage medium and equipment based on behavior tree Download PDFInfo
- Publication number
- CN116519316A CN116519316A CN202210071963.5A CN202210071963A CN116519316A CN 116519316 A CN116519316 A CN 116519316A CN 202210071963 A CN202210071963 A CN 202210071963A CN 116519316 A CN116519316 A CN 116519316A
- Authority
- CN
- China
- Prior art keywords
- test
- behavior
- node
- atomic
- scene
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010998 test method Methods 0.000 title claims abstract description 15
- 238000012360 testing method Methods 0.000 claims abstract description 276
- 230000009471 action Effects 0.000 claims abstract description 268
- 238000000034 method Methods 0.000 claims abstract description 34
- 230000006399 behavior Effects 0.000 claims description 218
- 239000002243 precursor Substances 0.000 claims description 31
- 230000007704 transition Effects 0.000 claims description 25
- 239000011159 matrix material Substances 0.000 claims description 22
- 230000008569 process Effects 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 10
- 230000003542 behavioural effect Effects 0.000 claims description 7
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 238000012544 monitoring process Methods 0.000 claims description 4
- 238000004088 simulation Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 7
- 230000008859 change Effects 0.000 description 5
- 238000010276 construction Methods 0.000 description 3
- SAZUGELZHZOXHB-UHFFFAOYSA-N acecarbromal Chemical compound CCC(Br)(CC)C(=O)NC(=O)NC(C)=O SAZUGELZHZOXHB-UHFFFAOYSA-N 0.000 description 2
- 238000004873 anchoring Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012423 maintenance Methods 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
- 238000012545 processing Methods 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/007—Wheeled or endless-tracked vehicles
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an automatic driving test method, device, storage medium and equipment based on a behavior tree, and relates to the field of automatic driving. The method comprises the following steps: acquiring a test map, constructing a test scene based on the test map, selecting a dynamic target, and acquiring an atomic action set of the dynamic target and a behavior track of the dynamic target in each test sub-scene included in the test scene; decomposing the behavior track into at least one atomic action based on the atomic action set, and determining a time sequence logic relationship of the atomic action; combining the time sequence logic relations in each test sub-scene, and creating a corresponding behavior tree of the test scene based on the combined result; and running the behavior tree to obtain an automatic driving test result of the tested object. The invention realizes automatic driving test in non-solidified and richer and more various test scenes.
Description
Technical Field
The present invention relates to the field of autopilot, and in particular, to an autopilot test method, apparatus, storage medium and device based on a behavior tree.
Background
In recent years, the safety of automatic driving is receiving a great deal of attention due to the occurrence of information about casualties caused by more and more intelligent unmanned systems.
International standard ISO 21448 SOTIF (security of intended functions) states that it is necessary to generate many scenarios in the initial development phase of security testing to enable the identification of all possible scenarios. Therefore, constructing a diversified mass test scenario is a core problem of automatic driving test research. In autopilot testing, there are generally two methods to generate a test scenario, namely real road testing and simulator-based simulation testing. The real road test has high cost and low efficiency, and the simulator-based simulation test can solve the problems and generate a virtual test scene. There are many autopilot simulation test platforms in the related art, such as Carla, airSim, lgsvl, apollo.
However, the test scenario used by the autopilot simulation test platform in the related art is fixedly generated, and the test scenario is single and not rich and diverse enough.
Disclosure of Invention
Aiming at the problems of fixed generation and insufficient diversity of test scenes, the invention provides an automatic driving test method, an automatic driving test device, a storage medium and automatic driving test equipment based on a behavior tree.
The technical scheme of the invention comprises the following steps:
an automatic driving test method based on a behavior tree, characterized in that the method comprises the following steps:
acquiring a test map, constructing a test scene based on the test map, and selecting a dynamic target, and acquiring an atomic action set of the dynamic target and a behavior track of the dynamic target in each test sub-scene included in the test scene, wherein the dynamic target comprises a tested object and a dynamic element;
decomposing the behavior track into at least one atomic action based on the atomic action set, and determining a time sequence logic relationship of the atomic action;
combining the time sequence logic relations in each test sub-scene, and creating a corresponding behavior tree of the test scene based on the combined result;
and running the behavior tree to obtain an automatic driving test result of the tested object.
Optionally, the sequential logic relationship comprises: at least one of atomic motion series, atomic motion parallel, and atomic motion repetition.
Optionally, said combining said sequential logic relationship in each of said test sub-scenarios includes:
combining the same dynamic target in each test sub-scene;
And combining the sequential logic relations of the atomic actions of different dynamic targets in each test sub-scene to obtain the combined result.
Optionally, the creating the corresponding behavior tree of the test scene includes:
marking behavior directed identifiers among the atomic actions based on the time sequence logic relationship, and generating a behavior directed graph according to the behavior directed identifiers;
constructing an adjacency matrix of the behavior directed graph by judging whether each atomic action in the behavior directed graph has a subsequent node;
creating a behavior tree corresponding to the test scene based on the adjacency matrix and behavior tree creation conditions, wherein the behavior tree creation conditions comprise:
adding a sequential parent node before each of the atomic actions;
if one atomic action has only one successor node, taking the successor node as a child node of a sequential parent node of the atomic action;
if one atomic action has a plurality of successor nodes, each successor node is used as a child node of a parallel father node, and the parallel father node is used as a child node of a sequential father node of the atomic action;
if one atomic action has a plurality of precursor nodes, taking the states of all the precursor nodes as a trigger event, taking the trigger event as a new precursor node of the atomic action, taking the new precursor node as a child node of a root node, and forming a new subtree by the new precursor node, the atomic action and a subsequent node of the atomic action;
If one of the atomic actions has no precursor node, the atomic action is used as a child node of a parallel node, and the parallel node is used as a root node of the behavior tree.
Optionally, the running the behavior tree includes:
establishing and initializing a behavior sub-tree, and starting a timer and a monitor of the behavior sub-tree, wherein the timer is used for recording the running time of the behavior tree, and the monitor is used for monitoring the abnormal condition of the behavior tree in the running process;
traversing each node of the behavior tree by the preamble, operating the node corresponding to the atomic action, and judging whether the operation behavior of the node meets the completion condition of the atomic action corresponding to the node so as to obtain the operation state of the node.
Optionally, the running the behavior tree further includes:
establishing a judging subtree, wherein the judging subtree is used for judging and evaluating the running result of the behavior tree;
the decision subtree is run with all nodes of the behavior tree traversed.
Optionally, the method further comprises:
constructing a transition scene and a behavior track of the dynamic target in the transition scene based on any two adjacent test sub-scenes and the behavior track of the dynamic target in the test sub-scenes; wherein the test scene includes the test sub-scene and the transition scene.
An automatic driving test device based on a behavior tree, comprising:
the setting module is used for acquiring a test map, constructing a test scene based on the test map, selecting a dynamic target, and acquiring an atomic action set of the dynamic target and a behavior track of the dynamic target in each test sub-scene included in the test scene, wherein the dynamic target comprises a tested object and a dynamic element;
the generation module is used for decomposing the behavior track into at least one atomic action based on the atomic action set and determining a time sequence logic relationship of the atomic action;
the creation module is used for combining the time sequence logic relations in each test sub-scene and creating a corresponding behavior tree of the test scene based on the combination result;
and the operation module is used for operating the behavior tree to obtain an automatic driving test result of the tested object.
A computer device comprising a memory and a processor, the memory having stored therein a computer program that is loaded and executed by the processor to implement the behavioral tree-based autopilot test method described above.
A computer readable storage medium having stored thereon a computer program which when executed by a processor implements an automated driving test method based on a behavior tree as described above.
A computer program product which, when run on a computer device, causes the computer device to perform an automated driving test method based on a behavior tree as described above.
Compared with the prior art, the invention has the following advantages:
1. aiming at the application of automatic driving simulation test, a behavior control scheme for realizing the tested object and the dynamic element based on a behavior tree is provided, and a large number of rich and various dynamic test scenes can be generated through the serial-parallel combination of the atomic actions of the tested object and the dynamic element in the process of generating the behavior tree;
2. the description of the time sequence logic relationship is introduced in the combination of the atomic actions, the atomic actions of the measured object and the dynamic elements are combined in time and space, a new view angle for describing different individual behaviors in time and space based on the time sequence logic relationship is provided, the behavior description is clearer and more concise, and the behavior interaction between different subjects is conveniently embodied;
3. Dynamic elements dynamically appear in the test sub-scenes, so that the number of the dynamic elements in different test sub-scenes can be flexibly adjusted, and the dynamic elements can be flexibly added and deleted;
4. the test scenes are combined through the plurality of test sub-scenes, so that flexible combination and construction of the test scenes can be realized, and the method is suitable for different test tasks;
5. the transition scenes are added among different test sub-scenes appropriately, so that the switching among the different test sub-scenes can be smoothed, the actual test is more fit, and the running consistency of the tested object in the different test sub-scenes is ensured.
Drawings
FIG. 1 is a flow chart of an automatic driving test method based on a behavior tree.
Fig. 2 is a schematic diagram of a test sub-scenario provided in an embodiment of the present application.
Fig. 3 is a schematic diagram of another test sub-scenario provided in an embodiment of the present application.
FIG. 4 is a schematic diagram of a sequential logic relationship provided in an embodiment of the present application.
FIG. 5 provides a behavioral directed graph according to an embodiment of the present application.
FIG. 6 provides another behavioral directed graph in accordance with an embodiment of the present application.
FIG. 7 is a schematic diagram of a behavior tree creation process according to an embodiment of the present application.
FIG. 8 is a schematic diagram of a behavior tree provided in an embodiment of the present application.
Fig. 9 is a block diagram of an automatic driving test equipment based on a behavior tree.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is apparent that the embodiments described are merely specific embodiments of the present invention, and not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.
The automatic driving test method of the present invention, as shown in fig. 1, includes the following steps (steps 110 to 140).
Step 110: acquiring a test map, constructing a test scene based on the test map, selecting a dynamic target, and acquiring an atomic action set of the dynamic target and a behavior track of the dynamic target in each test sub-scene included in the test scene.
In this step, the present invention sets the test sub-scene, the dynamic target, and the atomic action set constituting the dynamic target and the behavior trace of the dynamic target, respectively, based on the acquired test map, including the following sub-steps (111 to 115).
Step 111: and obtaining a test map.
For autopilot simulation testing, the test map acquired includes, but is not limited to: lane information (e.g., forward lane, reverse lane, forward lane, and follow-up lane, etc.), intersection information (e.g., crossroads, t-junctions, roundabout, etc.), viaducts, highways, traffic lights, etc.
Step 112: and constructing a test scene based on the test map.
The test scene may include at least one test sub-scene, which should be adapted to the test map. For example, assuming that the test map includes lane information and intersection information, a test sub-scene constructed based on the test map may include straight lanes, curves, intersections, and the like; assuming that the test map includes a highway, a test sub-scene constructed based on the test map may include: straight lanes, curves, etc.
Step 113: dynamic targets are selected based on the test scenario.
In an example, the dynamic target may include only the object under test (test carriage).
In yet another example, the dynamic target includes dynamic elements in addition to the object under test. The dynamic elements are selected to be suitable for the test scene, for example, in the case that the test scene comprises an intersection, the dynamic elements can comprise vehicles, pedestrians, traffic lights, weather and the like; in the case where the test scene includes an expressway, the dynamic elements may include traffic flow, weather, and the like.
It should be noted that the tested object usually continuously appears in the testing process, so that each test sub-scene of the test scene contains the tested object; however, the dynamic element is other elements that dynamically appear during the test except for the object under test, which may continuously appear during the test or may appear during a certain period of time during the test, so that a certain test sub-scene of the test scene may or may not contain the dynamic element, and the dynamic elements in the respective test sub-scenes of the test scene may or may not be the same.
Illustratively, the test map includes lane information and intersection information, and the test scene includes two test sub-scenes, namely a test sub-scene 1 as shown in fig. 2 and a test sub-scene 2 as shown in fig. 3. As shown in fig. 2 and 3, the test sub-scene 1 and the test sub-scene 2 each include an object to be tested: white car. As shown in fig. 2, the test sub-scene 1 includes two parallel straight lanes, and dynamic elements: black car, sunny day (not shown in fig. 2); as shown in fig. 3, the test sub-scene 2 includes two parallel left turn lanes, and dynamic elements: pedestrians, sunny days (not shown in fig. 3), traffic lights (not shown in fig. 3).
Step 114: and acquiring an atomic action set of the dynamic target.
The atomic action set of the present invention refers to a set composed of all possible atomic actions performed, and an atomic action refers to a minimum action unit.
For autopilot simulation testing, the atomic action set includes, but is not limited to: the atomic actions of test vehicles and vehicles (such as keeping vehicle speed, parking, braking, reversing, changing lanes, turning U-bend, straight running, turning left, turning right, keeping lane, anchoring, overtaking, rear-end collision, plugging, etc.), the atomic actions of traffic flows (such as vehicle speed, number of vehicles in traffic flows, intervals between vehicles, etc.), the atomic actions of pedestrians and pedestrian flows (such as crossing roads, walking along roads, etc.), the atomic actions of weather (such as sunny days, rainy days, snowy days, foggy days, etc.), the atomic actions of traffic lights (such as red lights, yellow lights, green lights, etc.).
In the present invention, the atomic action set of the dynamic target may be adapted to the selected test map, for example, in the case where the test map includes lane information (e.g., forward lane, reverse lane, precursor lane, and subsequent lane, etc.), intersection information (e.g., intersection, t-intersection, etc.), traffic lights, etc., so that the atomic action set may include: atomic actions of test vehicle, test traffic, traffic; atomic actions of pedestrians and pedestrian flows; atomic motion of traffic lights.
Step 115: and acquiring the behavior track of the dynamic target in each test sub-scene.
In the invention, the preset behavior track refers to the behavior track of the dynamic target in the test process. It should be understood that, the preset behavior trace herein refers to a running route (similar to navigation in a vehicle-mounted map) formed after setting a start point and an end point of a dynamic target and passing a task point in a testing process, and does not include parameter information such as a movement speed, a movement distance and the like of a tested object and a dynamic element.
For example, as shown in fig. 2, for the test sub-scenario 1, the preset behavior trace of the object under test (white car) includes: the white vehicle moves straight on the right lane; the preset behavior trace of the dynamic element (black car) includes: the black vehicle is firstly in straight running on the left lane, then in lane changing to the right lane, and then in straight running on the right lane for a certain distance after the lane changing is finished. As another example, as shown in fig. 3, for the test sub-scene 2, the preset behavior trace of the object under test (white car) includes: the white vehicle firstly moves straight and then turns left at the intersection; the preset behavior trace of the dynamic element (pedestrian) includes: the pedestrian waits at the intersection and then traverses the road.
Likewise, in the present application, the preset behavior trace may be adapted to the selected test map or to the structure of the test map. For example, the test map may include a straight lane and an intersection, and when the intersection is connected after the test map is the straight lane, since the right turn cannot be performed in the straight lane, the preset behavior trace should include: straight and then turn right at the intersection.
It should be noted that, the execution sequence of the step 114 and the step 115 is not limited in this application, and the step 114 may be executed before the step 115, may be executed after the step 115, or may be executed simultaneously with the step 115.
Step 120: based on the set of atomic actions, the behavior trace is decomposed into at least one atomic action, and a temporal logic relationship of the atomic actions is determined.
In the step, the invention constructs the time sequence logic relation of the atomic actions of different dynamic targets in each test sub-scene based on the obtained atomic action set and the action track of the dynamic target. The atomic action time sequence logic relation comprises: at least one of atomic motion series, atomic motion parallel, and atomic motion repetition. The serial of atomic actions means that two atomic actions are performed sequentially in time, for example, as shown in fig. 4, atomic action 1 starts to perform atomic action 2 after the execution is completed; the parallel of the atomic actions means that there is an intersection between the execution times of two atomic actions, for example, as shown in fig. 4, atomic action 1 and atomic action 2 start to be executed simultaneously, or atomic action 1 starts to execute atomic action 2 after a period of time is executed; the repetition of the atomic operation means that the atomic operation is repeatedly executed, and for example, as shown in fig. 4, the atomic operation 1 is executed again after the atomic operation 1 is executed.
First, the computer device may decompose a preset behavior trace of the dynamic target into a plurality of atomic actions based on the atomic actions included in the atomic action set. For example, as shown in fig. 2, for the test sub-scenario 1, the atomic actions that can be decomposed to obtain a white car based on the preset behavior trace of the object under test (white car) include: the atomic actions of the black car which can be decomposed based on the preset action track of the dynamic element (black car) comprise the following steps: straight running (keeping the vehicle speed), right lane changing, straight running, braking and stopping; as shown in fig. 3, for the test sub-scene 2, the atomic actions that can be decomposed to obtain a white car based on the preset behavior trace of the object under test (white car) include: the atomic actions of the pedestrians, which can be decomposed based on the preset behavior track of the dynamic element (the pedestrians), comprise the following steps: waiting and traversing the road (fixed speed).
And then, in each test sub-scene, determining the time sequence logic relation of the atomic actions of each dynamic target according to the action track decomposition result.
As shown in fig. 2, for test sub-scenario 1, the determined sequential logic relationship of atomic actions of the object under test (blue car) is: repeating the atomic operation of "straight running (keeping vehicle speed)", the determined sequential logic relationship of the atomic operation of the dynamic element (red car) is: serial of atomic actions such as "straight running (keeping vehicle speed), lane change to right, straight running, braking, and stopping"; as shown in fig. 3, for the test sub-scenario 2, the determined sequential logic relationship of the atomic actions of the object under test (blue car) is: the serial of the atomic actions of straight and left turn, and the determined time sequence logic relationship of the atomic actions of the dynamic element (pedestrian) is as follows: serial of these atomic actions "wait, traverse the road (fixed speed)".
Step 130: and combining the time sequence logic relations in each test sub-scene, and creating a corresponding behavior tree of the test scene based on the combined result.
Wherein, combining the sequential logic relationship in each test sub-scenario includes step 131 and step 132; based on the combined result, a corresponding behavior tree of the test scene is created, including steps 133 to 135.
Step 131: and combining the same dynamic target in each test sub-scene.
The tested object usually continuously appears in the test process, so that the tested object appears in each test sub-scene of the test scene, that is, the same tested object usually appears in a plurality of test sub-scenes. For example, as shown in fig. 2 and 3, when the object to be tested (white car) appears in both the test sub-scene 1 and the test sub-scene 2, the object to be tested (white car) in the test sub-scene 1 and the test sub-scene 2 need to be merged.
The dynamic elements dynamically appear in the test process, so that the same dynamic element can only appear in one test sub-scene of the test scene or can also appear in a plurality of test sub-scenes of the test scene. If the same dynamic element appears in multiple test sub-scenes, the dynamic elements in the multiple test sub-scenes need to be combined.
Step 132: and combining the sequential logic relations of the atomic actions of different dynamic targets in each test sub-scene to obtain a combined result.
The sequential logic relations of the atomic actions of different tested objects and dynamic elements in each test sub-scene can be combined to form a comprehensive combination of sequential logic relations of multiple tested objects and multiple dynamic elements in multiple test sub-scenes. When the time sequence logic relation of the atomic actions in each test sub-scene is combined, the dynamic elements which are newly added in different test sub-scenes are added, and the dynamic elements which do not appear are deleted.
For example, the sequential logic relationship of atomic actions based on the combination of test sub-scenario 1 as shown in FIG. 2 and test sub-scenario 2 as shown in FIG. 3 is: in the test sub-scene 1, the straight running (keeping speed) of the object to be tested (white car) and the straight running (keeping speed) of the dynamic element (black car) are parallel atomic actions, the straight running speed of the black car is slightly larger than the straight running speed of the white car, and the straight running (keeping speed) of the white car and the rightward lane changing, straight running, braking and stopping of the black car are parallel atomic actions; in the test sub-scene 2, the straight-through of the white vehicle and the right turn of the white vehicle are serial atomic actions, the waiting of the right turn and dynamic elements of the white vehicle (pedestrians need to be eliminated and pedestrians added at this time) are parallel atomic actions, and the waiting of the pedestrians and the crossing of the road (fixed speed) of the pedestrians are serial atomic actions.
In the application, because the map information and dynamic elements of different test sub-scenes may be different, in order to smooth the interface display effect when the different test sub-scenes are switched, a transition scene can be added between the different test sub-scenes, so that the combined atomic actions more conform to the actual scene. Based on this, in one example, the method provided by the present invention further includes: constructing a behavior track of a transition scene and a dynamic target in the transition scene based on the behavior tracks of any two adjacent test sub-scenes and the dynamic target in the test sub-scenes; the test scene comprises a test sub-scene and a transition scene. Optionally, the transition scene includes, but is not limited to, a map transition scene and a behavior transition scene, wherein the map transition scene refers to adding a new map information between two map information, and the behavior transition scene refers to adjusting the speed, lane, position and the like of the measured object and the dynamic element.
For example, based on test sub-scenario 1 as shown in fig. 2 and based on test sub-scenario 2 as shown in fig. 3, a transition scenario 3 may be added between test sub-scenario 1 and test sub-scenario 2, and the map transition scenario of transition scenario 3 may be: the straight road section, the behavior transition scene may be: the tested object (white car) changes the lane to bypass the stopped dynamic element (black car) and accelerates to a fixed vehicle speed when the white car is running in the test sub-scene 2.
Step 133: and marking behavior directed identifiers among the atomic actions based on the time sequence logic relationship, and generating a behavior directed graph according to the behavior directed identifiers.
Aiming at the difference of time sequence logic relations, the invention marks the behavior directional marks among the atomic actions by the following modes:
for the serial atomic actions, an atomic action A with a previous start execution time is set as a precursor node, an atomic action B with a subsequent start execution time is set as a subsequent node of the atomic action A, the end execution time of the precursor node is the start execution time of the subsequent node, and the behavior directional mark points to the atomic action B from the atomic action A.
For parallel atomic actions, if the starting execution time of two or more atomic actions (for example, atomic action B and atomic action C) is the same, then the atomic action B and the atomic action C are set as parallel nodes in the behavior directed graph; if the start execution time of each of the plurality of parallel atomic actions (for example, atomic action B and atomic action C) is the end execution time of a certain atomic action (for example, atomic action a), then the atomic action B and atomic action C are simultaneously used as successor nodes of the atomic action a, and the behavior directional identifier points to the atomic action B and the atomic action C respectively by the atomic action a.
For repeating the atomic actions, taking the atomic action A as an example, setting the atomic action A as a successor node of the atomic action A, wherein the ending execution time of the atomic action A is the starting execution time of executing the atomic action A again, and the atomic action A is pointed to by the atomic action A in a direction.
After the behavior directed identifiers among the atomic actions are obtained, one-to-one correspondence is carried out according to the time sequence logic relationship, so that a behavior directed graph of the time sequence logic relationship is generated.
Illustratively, a behavior directed graph as shown in fig. 5 may be obtained based on test sub-scenario 1 as shown in fig. 2 and based on test sub-scenario 2 as shown in fig. 3, and based on transition scenario 3 added between test sub-scenario 1 and test sub-scenario 2 as described above. Of course, since the initialization setting needs to be performed on the test scene before the behavior scheduling of the object to be tested and the dynamic element is performed through the behavior tree, as shown in fig. 5, the test scene initialization setting may be performed as the first atomic action to be added to the behavior directed graph. For other description of the atomic action of the test scenario initialization setting test scenario, please refer to the following embodiments, which are not repeated here.
Step 134: and constructing an adjacency matrix of the behavior directed graph by judging whether each atomic action in the behavior directed graph has a successor node.
The row of the adjacency matrix represents the atomic actions of each node in the row-directed graph, the column of the adjacency matrix represents whether the atomic actions corresponding to the row of the adjacency matrix have successor nodes, for example, 0 represents that the atomic actions of the row have no corresponding successor nodes, and 1 represents that the atomic actions of the row have corresponding successor nodes.
Illustratively, by way of a behavioral directed graph as shown in FIG. 6, an adjacency matrix as shown in Table 1 may be generated.
TABLE 1 adjacency matrix
Start to | Atomic motion 1 | Atomic action 2 | Atomic action 3 | Atomic action 4 | Atomic action 5 | Ending |
Atomic motion 1 | 0 | 1 | 1 | 0 | 0 | 0 |
Atomic action 2 | 0 | 0 | 0 | 0 | 0 | 0 |
Atomic action 3 | 0 | 0 | 0 | 0 | 1 | 0 |
Atomic action 4 | 0 | 0 | 0 | 0 | 1 | 0 |
Atomic action 5 | 0 | 0 | 0 | 0 | 0 | 1 |
Ending | 0 | 0 | 0 | 0 | 0 | 0 |
Illustratively, by way of a behavioral directed graph as shown in FIG. 5, an adjacency matrix as shown in Table 2 may be generated.
TABLE 2 adjacency matrix
Step 135: and creating a behavior tree corresponding to the test scene based on the adjacency matrix and the behavior tree creation condition.
The behavior directed graph is used for describing behaviors of the tested object and the dynamic element, and a behavior tree of the tested object and the dynamic element, namely a corresponding behavior tree of the test scene, can be created based on the adjacency matrix corresponding to the behavior directed graph. In the creation of the behavior tree, the following behavior tree creation conditions need to be followed:
(1) Each atomic action shown in the adjacency matrix is preceded by a Sequence (Sequence) parent node;
(2) In the atomic actions shown in the adjacency matrix, if a certain atomic action a has only one successor node (taking an atomic action B as an example), the successor node (atomic action B) is taken as a child node of a sequential parent node of the atomic action a;
(3) If a certain atomic action a has a plurality of successor nodes (taking an atomic action B and an atomic action C as an example), all the successor nodes (the atomic action B and the atomic action C) are taken as child nodes of a Parallel (Parallel) parent node, and the Parallel parent node is taken as a child node of a sequential parent node of the atomic action a;
(4) In the atomic actions shown in the adjacent matrix, if a certain atomic action E has a plurality of precursor nodes (taking an atomic action C and an atomic action D as examples), taking the states of all the precursor nodes (the atomic action C and the atomic action D) as a trigger event, taking the trigger event as a new precursor node of the atomic action E, and taking the atomic action E, the successor nodes and the new precursor node as a child node of a root node;
(5) If one atomic action E has no precursor node, the atomic action E is taken as a child node of a parallel node, and the parallel node is the root node of the behavior tree.
Note that since the sequential parent node is added before each atomic action in condition (1), the atomic actions described in conditions (2) - (5) should actually be the atomic actions to which the sequential parent node is added. The creation process of the behavior tree may be performed in the order of the conditions (1), (2), (3), (4), and (5), may be performed in the order of the conditions (5), (4), (3), (2), and (1), and may be performed in the order of the conditions (1), (3), (5), (2), and (4), and the invention is not limited thereto.
For example, please refer to fig. 7, which illustrates a process of creating a behavior tree based on the adjacency matrix shown in table 1. As shown in fig. 7 (a), the sequential parent node may be added first before each atomic action shown in table 1; then, as shown in fig. 7 (b), regarding the atomic actions in which the successor node exists, the successor node is taken as the child node (such as atomic action 5 and end) of the sequential parent node of the atomic action (the atomic action described herein is actually the atomic action to which the sequential parent node has been added), if there are multiple successor nodes, the multiple successor nodes are taken as the child nodes of one parallel parent node, and the parallel parent node is taken as the child node (such as atomic action 1, atomic action 2, and atomic action 3) of the sequential parent node of the atomic action; next, as shown in fig. 7 (c), regarding the atomic actions with multiple precursor nodes, the multiple precursor nodes are used as a trigger event, the trigger event is used as a new precursor node of the atomic action, then the atomic action and its subsequent nodes (atomic action 5 and end) and the new precursor node form a new subtree, and are used as a child node (such as atomic action 3, atomic action 4 and atomic action 5) of the root node; in addition, as shown in fig. 7 (c), for the atomic action without any precursor node, the atomic action is regarded as a child node (e.g., a start) of one parallel node (root node), and in fig. 7 (c), a behavior tree is created based on the adjacency matrix shown in table 1.
Illustratively, the same creation process as that of the behavior tree shown in fig. 7 may be employed to create a behavior tree of the object under test and the dynamic element based on the adjacency matrix shown in table 2, the created behavior tree being as shown in fig. 8.
Step 140: and running the behavior tree to obtain an automatic driving test result of the tested object.
In the present invention, step 140 includes the following steps 141 and 142.
Step 141: and establishing and initializing a behavior subtree, and starting a timer and a monitor of the behavior subtree.
When running the behavior tree, it is necessary to build a behavior subtree, and call an initialization function of the behavior subtree to initialize the behavior subtree. After initializing the behavior sub-tree, a timer of the behavior sub-tree and a listener of the behavior sub-tree are started. The timer of the behavior subtree is used for recording the running time of the whole behavior tree; the monitor of the behavior subtree is used for monitoring abnormal conditions of the whole behavior tree in the running process, such as the conditions that a test task cannot be performed, memory leaks or other system service anomalies are caused by connection failure of a simulation engine, collision or out of control of a tested object.
Step 142: and traversing each node of the behavior tree in advance, operating the node corresponding to the atomic action, and judging whether the operation behavior of the node meets the completion condition of the atomic action corresponding to the node so as to obtain the operation state of the node.
The preamble traversal refers to: firstly, accessing a root node, then traversing a left subtree, and finally traversing a right subtree; when traversing the left and right subtrees, the root node is still accessed first, then the left subtree is traversed, and finally the right subtree is traversed. Therefore, in the present application, the start node in the behavior tree is operated first, and then the node corresponding to the original action 1 in the behavior tree is operated.
And running the node corresponding to the atomic action 1, namely setting a test scene. The setting of the test scene comprises the following steps: and importing a test map, and generating a tested object and dynamic elements in the test map. Wherein, elements such as traffic lights, weather and the like in the dynamic elements are introduced when the atomic action of setting the test scene is executed, and the time of red lights, yellow lights, green lights, weather and the like of the traffic lights can be set. The elements such as vehicles, pedestrians and the like in the dynamic elements can be uniformly rendered when the test scene is set, and can be rendered again when the corresponding test sub-scene is operated. Taking the test sub-scene shown in fig. 2 and 3 as an example, since the test sub-scene shown in fig. 2 is operated first, when the node of the behavior tree, which is the set test scene, is operated, the rendered dynamic element is a black car, and subsequently when the test sub-scene 2 is operated or is about to be operated, the pedestrian is rendered again; of course, the present application does not exclude the possibility of rendering all dynamic elements when running the node of the behavior tree setting test scene.
As can be seen from the above description, the setting of the test scenario is the node that is operated first after the start node of the behavior tree, and after the setting of the test scenario, the behavior tree traverses the other nodes in turn according to the preamble traversal mode, so as to realize behavior control on the tested object and the dynamic element. In addition, in the present application, the setting test scenario may be implemented as one node of the behavior tree, such as the atomic action 1 in the behavior tree shown in fig. 8; the plurality of nodes may be realized as a plurality of nodes of the behavior tree, for example, each of the imported test map, the generated object to be tested, the generated dynamic element, and the like is one atomic action, and the atomic actions are executed in series or in parallel through the behavior tree.
And traversing the preamble, and if traversing to the node corresponding to the atomic action, running the node corresponding to the atomic action. For serial atomic actions, the nodes corresponding to the atomic actions are operated by using the sequential execution of the action tree according to the operation sequence; for parallel atomic actions, parallel execution of the behavior tree is used to run the node to which the atomic action corresponds. In the application, the specific steps of traversing and running the nodes corresponding to the atomic actions of the measured object and the dynamic element by the preamble are as follows:
Step 1421: and initializing parameters of the nodes when traversing the preamble to the nodes corresponding to the atomic actions.
When traversing to the node corresponding to the atomic actions of the measured object and the dynamic element, the node needs to be initialized with necessary parameters. The parameter initialization may be to initialize all parameters of the node, or may be to initialize some parameters of the node.
Step 1422: and updating the node state in real time, and setting the current state of the node as running.
Node states include, but are not limited to: run, success, failure. In the application, the update of the node state needs to perform behavior update of the measured object and the dynamic element, and then determine the update of the current state of the node based on the behavior update. This application is tested for autopilot simulation, behavior including, but not limited to: keeping the speed of a vehicle, parking, braking, reversing, changing the road, turning U-shaped bend, straight running, turning left and right, keeping the lane, anchoring, overtaking, rear-end collision, plugging and the like; the determination of the current state of a node may be based on, but is not limited to: whether the distance reaches the destination, whether the object to be tested collides, whether the traffic light is suitable for passing, etc.
For example, as shown in fig. 2, in the test sub-scenario 1, if an atomic motion obtained by decomposing the behavior of the object to be tested (white car) includes a straight motion (keeping the vehicle speed), when updating the current state of the node corresponding to the atomic motion, it is necessary to determine whether the vehicle speed of the straight motion of the white car is a prescribed vehicle speed; as shown in fig. 3, in the test sub-scene 2, if an atomic motion obtained by decomposing the behavior of the object to be tested (white car) includes a left turn, it is necessary to determine whether or not a pedestrian collides with the white car when the current state of the node corresponding to the atomic motion is updated.
Additionally, in one example, data of the current state of the node may be recorded as the current state of the node is updated. The data of the current state of the node includes, but is not limited to: the current speed, position (such as XY axis coordinates or longitude and latitude), lane, etc. of the measured object. Alternatively, the record of the data of the current state of the node may use a "Blackboard mechanism" (Blackboard), where the data recorded in the Blackboard is common data among behaviors in the behavior tree, and each node needs to read and write the data on the "Blackboard" during running, and may be implemented in the form of key value storage with global access rights.
For example, when executing a node corresponding to an atomic action, a behavior tree may be used to control the behavior of the object under test and the dynamic element, and one possible way to control the behavior through the behavior tree is shown below:
(1) Atomic actions describing the object under test and the dynamic element using speed and orientation.
The speed aspect needs to describe the numerical value of the speed of the measured object and the dynamic element and the speed change, so the content obtained by using the speed description atom action includes but is not limited to: acceleration, deceleration, vehicle speed maintenance, reverse gear, braking, parking, etc.; the aspect of orientation requires describing the orientation change of the measured object and the dynamic element in combination with road information, and thus the content obtained by using the orientation describing atomic action includes but is not limited to: lane changing to the left, lane changing to the right, lane keeping, left turn, right turn, turn around, straight, etc.
(2) The atomic actions related to the speed are set as a speed subtree, and the speed is controlled using the speed subtree.
The speed subtree includes a speed setting atomic action and a speed determining atomic action. The speed setting atomic actions are used for transmitting the target speed through the controller, and the controller calculates parameters such as a throttle, a brake and the like according to the current speed and the target speed and applies the parameters to the vehicle to control the speed of the vehicle; the speed determination atomic action obtains the speed of the vehicle from the data pool to determine whether the action is complete.
(3) An atomic action related to the orientation is set as a direction subtree, and travel is controlled using the direction subtree.
The direction subtree transmits the direction through the controller, acquires the target point from the data pool, and the controller plans the driving route according to the target direction and calculates the degree of the steering wheel at the same time; and then controlling the tested object and the dynamic element to run according to the planned running route by changing the degree of the steering wheel.
Step 1423: and when the behavior meets the completion condition of the atomic action, setting the current state of the node as successful.
When the node state is updated in real time, the behaviors of the measured object and the dynamic element need to be updated first, and when the behaviors meet the completion condition of the current atomic action, the current state of the node corresponding to the current atomic action is set as success, and the node corresponding to the atomic action is completed in operation.
Illustratively, as shown in fig. 2, the atomic actions obtained by decomposing the behavior of the dynamic element (black car) in the test sub-scene 1 include lane change to the right, and the completion condition of the atomic actions may be set to start straight running of the black car; when the node corresponding to the atomic motion is operated, the behavior of the black car needs to be acquired, if the behavior of the black car starts to turn to the right to change the lane, the behavior of the black car still does not meet the completion condition of the atomic motion of turning to the right, if the behavior of the black car starts to go straight, the behavior of the black car meets the completion condition of the atomic motion of turning to the right, at the moment, the current state of the node corresponding to the atomic motion can be set to be successful, and the operation of the node corresponding to the atomic motion of turning to the right is completed.
Step 1424: terminating the operation of the node and entering the operation of the next node according to the sequence of the preamble traversal.
When the node operation corresponding to the current atomic action is completed, the operation of the node may be terminated, and the next node of the behavior tree may be continuously traversed according to the sequence of the previous traversal, and then the operation of the next node is entered, that is, steps 1421 to 1424 are repeatedly executed.
After traversing all nodes of the behavioral tree through the above steps 141 to 142, test data may be generated and locally recorded and stored; of course, the test data may be subjected to processing such as derivation.
In one example, step 141 may simultaneously create a behavior sub-tree and a decision sub-tree when running the behavior tree, the decision sub-tree being used to decide and evaluate the running results of the behavior tree. Thus, step 140 of the present invention further comprises: establishing a decision sub-tree; in the case of traversing all nodes of the completion behavior tree, the decision subtree is run. After traversing all nodes of the behavior tree, the tested object completes all tasks in the test environment, and enters a decision subtree to decide and evaluate the operation result. The determining and evaluating of the operation result may be based on a test index defined by the test task, which specifically includes but is not limited to: whether red light is running, the number of red light running, whether overspeed, collision, deviation from the original driving route, overtime running, retrograde running, and the like.
In summary, the method decomposes the preset behavior trace of the object to be tested and the dynamic element into a plurality of atomic actions based on the atomic action set obtained by abstraction, then combines the atomic actions of the object to be tested and the dynamic element in time and space based on the sequential logic relationship (serial/parallel/repeated) of the atomic actions, can generate a large number of rich and various dynamic test scenes, and can provide a new view angle for describing different individual behaviors in time and space based on the sequential logic relationship relative to the traditional technology for controlling the behaviors of the object to be tested and the dynamic element by codes.
According to the invention, the test scenes are combined through the plurality of test sub-scenes, and the transition scenes are added between different test sub-scenes, so that not only can the flexible combination and construction of the test scenes be realized, so as to be suitable for different test tasks, but also the switching between different test sub-scenes can be smoothed, so that the test scenes are more suitable for actual tests, and the running consistency of the tested object in different test sub-scenes is ensured.
When the method is operated to the corresponding test sub-scene, the dynamic elements can be flexibly deleted, and compared with the prior art that the number, the appearance time and the like of the dynamic elements need to be set in advance in the form of codes, the method can enable the dynamic elements to dynamically appear in the test sub-scene.
Referring to fig. 9, a block diagram of an automatic driving test device based on a behavior tree according to an embodiment of the present invention is shown. The device can be a computer device or can be arranged in the computer device. As shown in fig. 9, the apparatus 900 includes: a setup module 910, a generation module 920, a creation module 930, and a run module 940.
The setting module 910 is configured to obtain a test map, construct a test scene based on the test map, and select a dynamic target, and obtain an atomic action set of the dynamic target and a behavior track of the dynamic target in each test sub-scene included in the test scene, where the dynamic target includes a tested object and a dynamic element;
A generating module 920, configured to decompose the behavior trace into at least one atomic action based on the atomic action set, and determine a sequential logic relationship of the atomic actions;
a creating module 930, configured to combine the sequential logic relationships in each of the test sub-scenarios, and create a corresponding behavior tree of the test scenario based on the combined result;
and the operation module 940 is used for operating the behavior tree to obtain an automatic driving test result of the tested object.
Optionally, the sequential logic relationship comprises: at least one of atomic motion series, atomic motion parallel, and atomic motion repetition.
Optionally, the creating module 930 includes a sequential logical relationship combining unit configured to:
combining the same dynamic target in each test sub-scene;
and combining the sequential logic relations of the atomic actions of different dynamic targets in each test sub-scene to obtain the combined result.
Optionally, the creating module 930 includes a behavior tree creating unit configured to:
marking behavior directed identifiers among the atomic actions based on the time sequence logic relationship, and generating a behavior directed graph according to the behavior directed identifiers;
Constructing an adjacency matrix of the behavior directed graph by judging whether each atomic action in the behavior directed graph has a subsequent node;
creating a behavior tree corresponding to the test scene based on the adjacency matrix and behavior tree creation conditions, wherein the behavior tree creation conditions comprise:
adding a sequential parent node prior to the atomic action;
if one atomic action has only one successor node, taking the successor node as a child node of a sequential parent node of the atomic action;
if one atomic action has a plurality of successor nodes, each successor node is used as a child node of a parallel father node, and the parallel father node is used as a child node of a sequential father node of the atomic action;
if one atomic action has a plurality of precursor nodes, taking the states of all the precursor nodes as a trigger event, taking the trigger event as a new precursor node of the atomic action, taking the new precursor node as a child node of a root node, and forming a new subtree by the new precursor node, the atomic action and a subsequent node of the atomic action;
if the atomic actions do not have the precursor nodes, the atomic actions are used as child nodes of a parallel node, and the parallel node is used as a behavior tree root node.
Optionally, the operation module 940 is configured to:
establishing and initializing a behavior sub-tree, and starting a timer and a monitor of the behavior sub-tree, wherein the timer is used for recording the running time of the behavior tree, and the monitor is used for monitoring the abnormal condition of the behavior tree in the running process;
traversing each node of the behavior tree by the preamble, operating the node corresponding to the atomic action, and judging whether the operation behavior of the node meets the completion condition of the atomic action corresponding to the node so as to obtain the operation state of the node.
Optionally, the operation module 940 is further configured to:
establishing a judging subtree, wherein the judging subtree is used for judging and evaluating the running result of the behavior tree;
the decision subtree is run with all nodes of the behavior tree traversed.
Optionally, the apparatus 900 further includes a transitional scene construction module 950 configured to: constructing a transition scene and a behavior track of the dynamic target in the transition scene based on any two adjacent test sub-scenes and the behavior track of the dynamic target in the test sub-scenes; wherein the test scene includes the test sub-scene and the transition scene.
For details of the specific implementation process, beneficial effects, etc. of the device module, please refer to the description of the above method embodiment, and the details are not repeated here.
In an exemplary embodiment, there is also provided a computer device comprising a memory and a processor, the memory having stored therein a computer program that is loaded and executed by the processor to implement the above-described behavior tree based autopilot test method.
In an exemplary embodiment, a computer readable storage medium is also provided, on which a computer program is stored which, when executed by a processor, implements an automatic driving test method based on a behavior tree as described above.
In an exemplary embodiment, a computer program product is also provided which, when run on a computer device, causes the computer device to perform an automatic driving test method based on a behavior tree as described above.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (10)
1. An automatic driving test method based on a behavior tree, characterized in that the method comprises the following steps:
acquiring a test map, constructing a test scene based on the test map, and selecting a dynamic target, and acquiring an atomic action set of the dynamic target and a behavior track of the dynamic target in each test sub-scene included in the test scene, wherein the dynamic target comprises a tested object and a dynamic element;
decomposing the behavior track into at least one atomic action based on the atomic action set, and determining a time sequence logic relationship of the atomic action;
combining the time sequence logic relations in each test sub-scene, and creating a corresponding behavior tree of the test scene based on the combined result;
and running the behavior tree to obtain an automatic driving test result of the tested object.
2. The method of claim 1, wherein the sequential logic relationship comprises: at least one of atomic motion series, atomic motion parallel, and atomic motion repetition.
3. The method of claim 1, wherein said combining said sequential logic relationship in each of said test sub-scenarios comprises:
Combining the same dynamic target in each test sub-scene;
and combining the sequential logic relations of the atomic actions of different dynamic targets in each test sub-scene to obtain the combined result.
4. The method of claim 1, wherein creating the corresponding behavioral tree of the test scenario comprises:
marking behavior directed identifiers among the atomic actions based on the time sequence logic relationship, and generating a behavior directed graph according to the behavior directed identifiers;
constructing an adjacency matrix of the behavior directed graph by judging whether each atomic action in the behavior directed graph has a subsequent node;
creating a behavior tree corresponding to the test scene based on the adjacency matrix and behavior tree creation conditions, wherein the behavior tree creation conditions comprise:
adding a sequential parent node before each of the atomic actions;
if one atomic action has only one successor node, taking the successor node as a child node of a sequential parent node of the atomic action;
if one atomic action has a plurality of successor nodes, each successor node is used as a child node of a parallel father node, and the parallel father node is used as a child node of a sequential father node of the atomic action;
If one atomic action has a plurality of precursor nodes, taking the states of all the precursor nodes as a trigger event, taking the trigger event as a new precursor node of the atomic action, taking the new precursor node as a child node of a root node, and forming a new subtree by the new precursor node, the atomic action and a subsequent node of the atomic action;
if one of the atomic actions has no precursor node, the atomic action is used as a child node of a parallel node, and the parallel node is used as a root node of the behavior tree.
5. The method of claim 1, wherein the running the behavior tree comprises:
establishing and initializing a behavior sub-tree, and starting a timer and a monitor of the behavior sub-tree, wherein the timer is used for recording the running time of the behavior tree, and the monitor is used for monitoring the abnormal condition of the behavior tree in the running process;
traversing each node of the behavior tree by the preamble, operating the node corresponding to the atomic action, and judging whether the operation behavior of the node meets the completion condition of the atomic action corresponding to the node so as to obtain the operation state of the node.
6. The method of claim 4, wherein the running the behavior tree further comprises:
establishing a judging subtree, wherein the judging subtree is used for judging and evaluating the running result of the behavior tree;
the decision subtree is run with all nodes of the behavior tree traversed.
7. The method of any one of claims 1 to 6, further comprising:
constructing a transition scene and a behavior track of the dynamic target in the transition scene based on any two adjacent test sub-scenes and the behavior track of the dynamic target in the test sub-scenes; wherein the test scene includes the test sub-scene and the transition scene.
8. An automatic driving test device based on a behavior tree, comprising:
the setting module is used for acquiring a test map, constructing a test scene based on the test map, selecting a dynamic target, and acquiring an atomic action set of the dynamic target and a behavior track of the dynamic target in each test sub-scene included in the test scene, wherein the dynamic target comprises a tested object and a dynamic element;
the generation module is used for decomposing the behavior track into at least one atomic action based on the atomic action set and determining a time sequence logic relationship of the atomic action;
The creation module is used for combining the time sequence logic relations in each test sub-scene and creating a corresponding behavior tree of the test scene based on the combination result;
and the operation module is used for operating the behavior tree to obtain an automatic driving test result of the tested object.
9. A computer device comprising a memory, in which a computer program is stored, and a processor arranged to run the computer program to perform the method of any of claims 1-7.
10. A storage medium having a computer program stored therein, wherein the computer program is arranged to perform the method of any of claims 1-7 when run.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210071963.5A CN116519316A (en) | 2022-01-21 | 2022-01-21 | Automatic driving test method, device, storage medium and equipment based on behavior tree |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210071963.5A CN116519316A (en) | 2022-01-21 | 2022-01-21 | Automatic driving test method, device, storage medium and equipment based on behavior tree |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116519316A true CN116519316A (en) | 2023-08-01 |
Family
ID=87403412
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210071963.5A Pending CN116519316A (en) | 2022-01-21 | 2022-01-21 | Automatic driving test method, device, storage medium and equipment based on behavior tree |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116519316A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117746714A (en) * | 2024-02-20 | 2024-03-22 | 成都运达科技股份有限公司 | Test method and system for simulated driving operation |
-
2022
- 2022-01-21 CN CN202210071963.5A patent/CN116519316A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117746714A (en) * | 2024-02-20 | 2024-03-22 | 成都运达科技股份有限公司 | Test method and system for simulated driving operation |
CN117746714B (en) * | 2024-02-20 | 2024-04-30 | 成都运达科技股份有限公司 | Test method and system for simulated driving operation |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112668153B (en) | Method, device and equipment for generating automatic driving simulation scene | |
US20230124864A1 (en) | Graph Representation Querying of Machine Learning Models for Traffic or Safety Rules | |
CN110162050B (en) | Travel control method and travel control system | |
CN111735639B (en) | Automatic driving scene minimum set generation method for intelligent networked automobile demonstration area | |
CN110162003B (en) | Driving active service system test verification platform and test method thereof | |
CN104760592B (en) | Self-propelled vehicle Decision Control method | |
CN113065257A (en) | Automatic generation method and device of test case, computer equipment and medium | |
CN112799964B (en) | Test scene generation method, device, equipment and storage medium | |
CN114168457A (en) | Method and system for generating test scene of automatic driving automobile | |
WO2009119006A1 (en) | Signalized intersection information acquiring device, signalized intersection information acquiring method, and signalized intersection information acquiring program | |
CN112035951A (en) | Simulation platform and simulation method for automatic driving algorithm verification | |
CN111832179B (en) | Unmanned vehicle test scene creation method and device | |
CN116519316A (en) | Automatic driving test method, device, storage medium and equipment based on behavior tree | |
CN115292816A (en) | Automatic driving test method, device, equipment and storage medium | |
Jiang et al. | A reinforcement learning benchmark for autonomous driving in general urban scenarios | |
Zhou et al. | A survey on autonomous driving system simulators | |
CN116384150A (en) | Test case generation method and system for automatic driving system | |
CN117413257A (en) | Method and system for testing driver assistance system for vehicle | |
CN114692289A (en) | Automatic driving algorithm testing method and related equipment | |
Kalandyk | Reinforcement learning in car control: A brief survey | |
Meftah et al. | A survey on autonomous vehicles simulators. | |
Pasini et al. | Vehicular traffic simulations: use of a games engine for video rendering and verification of an architecture model based on queuing systems | |
CN112904843B (en) | Automatic driving scene determining method and device, electronic equipment and storage medium | |
CN116306029B (en) | Automatic driving simulation test method based on Carla and ROS | |
van Orden | Cheating by segmentation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |