CN116360422A - Control method of automatic cruising intelligent obstacle avoidance trolley - Google Patents
Control method of automatic cruising intelligent obstacle avoidance trolley Download PDFInfo
- Publication number
- CN116360422A CN116360422A CN202310091539.1A CN202310091539A CN116360422A CN 116360422 A CN116360422 A CN 116360422A CN 202310091539 A CN202310091539 A CN 202310091539A CN 116360422 A CN116360422 A CN 116360422A
- Authority
- CN
- China
- Prior art keywords
- trolley
- intelligent
- line set
- preset
- turn
- 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
- 238000000034 method Methods 0.000 title claims abstract description 85
- 239000004576 sand Substances 0.000 claims abstract description 18
- 230000008569 process Effects 0.000 claims description 20
- 230000001419 dependent effect Effects 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000013135 deep learning Methods 0.000 claims description 5
- 230000001629 suppression Effects 0.000 claims description 3
- 238000011161 development Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 238000004891 communication Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012805 post-processing Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000004377 microelectronic Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0231—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means
- G05D1/0246—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means using a video camera in combination with image processing means
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Electromagnetism (AREA)
- Aviation & Aerospace Engineering (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
The invention belongs to the technical field of intelligent trolleys, and particularly relates to a control method of an automatic cruising intelligent obstacle avoidance trolley, which comprises the steps of placing the intelligent trolleys on a driving road of a traffic sand table and collecting real-time road condition images; the method comprises the following steps: s1, analyzing road condition images and identifying information of objects on a road; s2, estimating the distance between the intelligent trolley and each object with the height by combining the pre-stored real heights of the objects; s3, if a traffic signal lamp or a pedestrian in a red light state exists in the identified object and the distance between the traffic signal lamp or the pedestrian and the trolley is smaller than a corresponding distance threshold value, turning to S4, otherwise turning to S5; s4, executing a preset waiting strategy; s5, if the recognized object has an obstacle and the distance between the recognized object and the trolley is smaller than a preset safe avoidance distance, executing a preset obstacle avoidance driving strategy; otherwise, turning to S6; and S6, executing preset autonomous cruise driving. The method can improve the intelligent degree of the intelligent trolley and the strain capacity of the intelligent trolley on traffic events.
Description
Technical Field
The invention belongs to the technical field of intelligent trolleys, and particularly relates to a control method of an automatic cruising intelligent obstacle avoidance trolley.
Background
Along with the rapid development of computer and microelectronic technologies, the development of intelligent technologies is faster and faster, the intelligent degree is higher and higher, and the application range is greatly expanded. The intelligent trolley is used as the miniature of the intelligent automobile, the technologies of computer, communication, sensing and the like are intensively applied, and the intelligent mobile robot integrates the functions of environment sensing, planning decision, multi-level auxiliary driving and the like.
Through integrating above-mentioned technique, intelligent dolly can carry out the intelligent driving of certain degree on traffic sand table. However, the traditional intelligent trolley can only finish a single tracking task on a fixed track on a traffic sand table, has poor robustness, can not finish tasks such as cruising in more complex and changeable road scenes on the traffic sand table, and has very limited practical application range and more limitation in use; in addition, at present, the intelligent trolley generally adopts an infrared obstacle avoidance or ultrasonic obstacle avoidance as an obstacle avoidance scheme, so that the control scheme is single, the intelligent degree is not high, and more flexible obstacle avoidance cannot be performed. These have greatly limited the application and secondary development of intelligent carts.
Therefore, how to improve the intelligent degree of the intelligent trolley and the strain capacity of traffic incidents and to expand the practical application range of the intelligent trolley on the traffic sand table as much as possible becomes a current problem to be solved urgently.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides the control method of the automatic cruising intelligent obstacle avoidance trolley, which can improve the intelligent degree of the intelligent trolley and the strain capacity of traffic events and expand the practical application range of the intelligent trolley on a traffic sand table as much as possible.
In order to solve the technical problems, the invention adopts the following technical scheme:
a control method of an automatic cruising intelligent obstacle avoidance trolley comprises the steps of placing the intelligent trolley on a driving road of a traffic sand table, and collecting real-time road condition data through an acquisition unit of the intelligent trolley, wherein the road condition data comprises road condition images; then, the automatic cruising intelligent obstacle avoidance control of the intelligent trolley comprises the following steps:
s1, analyzing road condition images and identifying information of objects on a road; the information of the object comprises the type of the object, the type of the object comprises a lane line and an object with a height, and the object with the height comprises a traffic signal lamp, pedestrians and obstacles;
s2, estimating the distance between the intelligent trolley and each object with the height according to the information of the object identified in the S1 and the pre-stored real height of each object;
s3, carrying out driving strategy analysis according to the identified object information and the estimated distance between the intelligent trolley and the object with the height; if the traffic signal lamp in the red light state exists in the identified object and the distance between the traffic signal lamp and the trolley is smaller than the preset red light distance threshold value, or if the pedestrian exists and the distance between the traffic signal lamp and the trolley is smaller than the preset pedestrian distance threshold value, the step S4 is carried out, and otherwise, the step S5 is carried out;
s4, executing a preset waiting strategy;
s5, if the recognized object has an obstacle and the distance between the recognized object and the trolley is smaller than a preset safe avoidance distance, executing a preset obstacle avoidance driving strategy; otherwise, turning to S6;
and S6, executing preset autonomous cruise driving.
Preferably, S1 comprises:
s11, clipping the road condition image into the size of INPUT_W multiplied by INPUT_H pixel points, wherein INPUT_W is the width of the road condition image after clipping, and INPUT_H is the height of the road condition image after clipping;
s12, inputting the cut road condition image into a preset deep learning network for object recognition, decoding information under a preset confidence threshold conf_thresh, and obtaining the information of the object in the road condition image through a non-maximum suppression algorithm under a preset NMS threshold nms_thresh.
Preferably, in S1, the information of the object further includes position information and size information of the object; the position information of the object includes center point coordinates of the object; the size information of the object includes a pixel width and a pixel height of the object in the road condition image.
Preferably, in S1, the lane lines include a left straight line, a left turn line, a left right turn line, a right straight line, a right left turn line, and a right turn line; and identifying the lane lines according to the block areas with the preset size.
Preferably, the acquisition unit of the intelligent trolley comprises a camera; in S2, the estimation method of the distance dis between the intelligent trolley and the object with height is as follows:
dis= (true height×f)/height; wherein true height is the real height of the pre-stored object, f is the focal length of the camera of the intelligent trolley, and height is the height of the object detected in S1.
Preferably, in S4, the waiting policy includes:
if the distance between the trolley and the traffic signal lamp is smaller than a preset red light distance threshold value traffic_thresh, and the current signal lamp type is red light, the trolley is stopped for waiting, and the driving state before the signal lamp is recovered after being converted into a green light;
if the distance between the trolley and the pedestrian is smaller than the preset pedestrian distance threshold ped _thresh, the trolley is stopped for waiting, and the traveling state before the pedestrian is recovered after passing.
Preferably, in S5, the obstacle avoidance driving strategy includes:
s1, analyzing to obtain position information and size information of the obstacle;
if the coordinates of the central point of the obstacle
Judging that the obstacle is at the left side of the trolley, and controlling the trolley to turn right +.>
An angle, wherein maxTurnAngle represents the maximum steering angle supported by the trolley, and width represents the pixel width of the obstacle in the road condition image; after bypassing the obstacle, controlling the trolley to turn left by the same angle and return to the original running track;
if the coordinates of the central point of the obstacle
Judging that the obstacle is on the right side of the trolley and controllingThe car turns leftwards>
An angle, wherein maxTurnAngle represents the maximum steering angle supported by the cart; after the obstacle is bypassed, the trolley is controlled to turn right by the same angle and return to the original running track.
Preferably, in S6, the autonomous cruise driving strategy includes:
s6.1 divides the lane information identified in S1 into the following lane sets: left straight line: l (L) ls {l 1 ,l 2 ,…,l N1 Left turn line }: l (L) ll {l 1 ,l 2 ,…,l N2 Left-right turn line L lr {l 1 ,l 2 ,…,l N3 Straight line L on right side rs {{l 1 ,l 2 ,…,l N4 }, right left turn line L rl {l 1 ,l 2 ,…,l N5 Right side right turn line L rr {l 1 ,l 2 ,…,l N6 -a }; wherein the elements in each set are the coordinates of the central point of the block-shaped area of the lane line;
s6.2, judging that the current trolley is in a straight-going, left-turning or right-turning state according to the number of block area elements in each lane line set in the step S6.1, and obtaining an X coordinate lane_center X of the center position of the current driving lane;
s6.3 calculating an error value of the center position of the trolley and the current driving lane
Obtaining output quantity turnAngle (t) for controlling steering angle of the steering engine according to the position type PID; control using PD: turnAngle (t) =K P error(t)+K D [error(t)-error(t-1)]The method comprises the steps of carrying out a first treatment on the surface of the Wherein K is P And K D Respectively a proportional coefficient and a differential coefficient, wherein t represents time;
s6.4, obtaining the steering angle which is finally required to be adjusted by the trolley by adopting a dynamic window median filtering mode; the dynamic window median filtering includes: calculating turn angle values at the current n+1 times, taking the median value as an angle value to be adjusted of the trolley, wherein mectinangle=med [ turn angle (t), turn angle (t-1), …, turn angle (t-n) ],0<n; if the value exceeds the maximum steering angle maxTurnAngle supported by the trolley, the adjustment value is constrained to maxTurnAngle;
and S6.5, controlling the trolley to complete corresponding steering actions according to the calculated angle value which is finally required to be adjusted by the trolley in the step S6.4.
Preferably, in S6.2, the process of obtaining the X-coordinate lane_center X of the center position of the current driving lane includes:
comparing the number of elements in each lane line set, if the number of elements in the left straight line set is more than the left turning line set and the left right turning line set, and the number of elements in the right straight line set is more than the right left turning line set and the right turning line set, judging that the trolley is in a straight running state, and acquiring a left straight line set L ls Straight line set L on right side rs The method comprises the steps of carrying out a first treatment on the surface of the Will L ls And L rs Fitting to a straight line using the ordinate center Y of the center point as an independent variable and the abscissa center X of the center point as an independent variable X, respectively ls ＝k ls ×centerY ls +b ls ，centerX rs ＝k rs ×centerY rs +b rs The method comprises the steps of carrying out a first treatment on the surface of the Obtaining the coordinates of the position of the headstock at the moment
And the center position X coordinate of the current lane travel +.>
If the left turn line set has more elements than the left straight line set and the left right turn line set, and the right left turn line set has more elements than the right straight line set and the right turn line set, the vehicle is determined to be a trolleyIn a left turning state, acquiring a left turning line set L ll Set of right left turn lines L rl The method comprises the steps of carrying out a first treatment on the surface of the Will L ll And L rl Fitting to quadratic curves each having a central point ordinate center Y as an independent variable and a central point abscissa center X as a dependent variable X ll ＝k ll1 ×centerY 2 ll +k ll2 ×centerY ll +b ll ，centerX rl ＝k rl1 ×centerY 2 rl +k rl2 ×centerY rl +b rl The method comprises the steps of carrying out a first treatment on the surface of the Obtaining the coordinates of the position of the headstock at the moment as
Center position X coordinate of current lane travel
If the elements in the left-right turning line set are more than the left-left turning line set and the left straight line set, and the elements in the right-right turning line set are more than the right-left turning line set and the right straight line set, the trolley is judged to be in a right turning state, and a left-right turning line set L is acquired lr Right side right turn line set L rr The method comprises the steps of carrying out a first treatment on the surface of the Will L lr And L rr Fitting to quadratic curves each having a central point ordinate center Y as an independent variable and a central point abscissa center X as a dependent variable X lr ＝k lr1 ×centerY 2 lr +k lr2 ×centerY lr +b lr ，centerX rr ＝k rr1 ×centerY 2 rr +k rr2 ×centerY rr +b rr The method comprises the steps of carrying out a first treatment on the surface of the Obtaining the coordinates of the position of the headstock at the moment as
Center position X coordinate of current lane travel
Preferably, in S6.2, the fitting of the straight line or conic is done by least squares:
assuming that the current lane line set contains central point coordinate information of N block areas, the order of an equation to be fitted is M, and the following equation is satisfied:
k 1 ×centerY M +k 2 ×centerY M-1 +…+k M+1 ＝centerX；
wherein k is i Coefficients of the secondary terms in the corresponding equation, i e (1, 2 …, m+1);
the equivalent a·w=b is represented by a matrix as follows:
obtaining coefficient matrix W= (A) T A) -1 A T B, and obtaining M+1 coefficients k of M-order equation 1 ，k 2 ，…k M+1 ，
Finally, the equation of M-order curve or straight line is obtained by fitting: centrx=k 1 ×centerY M +k 2 ×centerY M-1 +…+k M+1 。
Compared with the prior art, the invention has the following beneficial effects:
1. by using the method, after the intelligent trolley is placed on a running road of a traffic sand table, real-time road condition images are acquired through the acquisition unit of the intelligent trolley, the road condition images are processed and analyzed, information of objects on the road condition images is extracted, and the distance between the intelligent trolley and the objects with heights is estimated.
Therefore, the current road condition of the intelligent trolley can be comprehensively and accurately known, and the corresponding driving strategy is selected. Specifically, if a traffic signal lamp in a red light state exists in the identified object and the distance between the traffic signal lamp and the trolley is smaller than a preset red light distance threshold value, or if a pedestrian exists and the distance between the traffic signal lamp and the trolley is smaller than a preset pedestrian distance threshold value, the intelligent trolley is indicated to have the risk of running the red light or colliding with the pedestrian, so that a predicted waiting strategy is executed, and the running rationality is ensured; if the risk of red light running or collision with pedestrians does not exist, but the recognized object has an obstacle (such as a cone barrel) and the distance between the recognized object and the trolley is smaller than the preset safe avoidance distance, the intelligent trolley can continue to run but needs to avoid the obstacle, so that a preset obstacle avoidance driving strategy is executed, and the running safety is ensured; if there is no risk, it is indicated that the intelligent vehicle can run normally, and thus preset autonomous cruise driving is performed.
Compared with the prior art, the intelligent trolley can complete the cruising task in a more complex semi-open traffic sand table environment, the intelligent trolley can realize functions of intelligent obstacle avoidance, automatic cruising and the like, the limitation that the intelligent trolley can only complete the tracking task in a fixed scene and the obstacle avoidance strategy is single is broken, and the intelligent degree of the intelligent trolley and the strain capacity of traffic events are improved.
In conclusion, the method can improve the intelligent degree of the intelligent trolley and the strain capacity of traffic events, and expands the practical application range of the intelligent trolley on a traffic sand table as much as possible.
2. According to the method, the road image in the traffic sand table is analyzed and processed in real time through the deep learning method and the post-processing algorithm, and compared with the method using infrared obstacle avoidance or ultrasonic obstacle avoidance as an obstacle avoidance scheme in the prior art, the method can more intelligently identify the obstacle to perform more flexible obstacle avoidance, so that the control mode of the intelligent trolley is more diversified, and the function expansion and secondary development are facilitated.
3. The method pre-stores the real heights of all objects on the sand table road, and the real distance between the intelligent trolley and the objects can be directly estimated through road condition images.
4. The obstacle avoidance driving strategy in the method can ensure that the trolley fully avoids the obstacle and can adjust the original driving track in time after the obstacle is avoided.
5. The method also provides a complete autonomous cruise driving strategy, and can ensure that the trolley is positioned at an accurate position as far as possible in the driving process. The autonomous driving strategy is completely based on actual data of each lane line set on the real-time road condition image, has clear logic and simple process, and can ensure the effectiveness and consistency of the autonomous cruise driving process.
Drawings
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings, in which:
FIG. 1 is a flow chart of an embodiment of the present invention;
fig. 2 is a schematic diagram of a measurement process of a focal length f of a camera in an embodiment;
FIG. 3 is a schematic diagram illustrating a process of obstacle avoidance driving strategy in an embodiment;
fig. 4 is a schematic diagram of a process of an autonomous cruise driving strategy in an embodiment.
Detailed Description
The following is a further detailed description of the embodiments:
examples:
to facilitate better practice of the present method by those skilled in the art, the following examples are set forth. In this embodiment, the upper computer of the intelligent trolley uses jetson nano microcomputer produced by inflight for data communication and image processing, and after the port and the baud rate are set, the serial port is used for communication with the stm32 lower computer to issue a motion control instruction for the intelligent trolley. The intelligent trolley is provided with the drive-free USB wide-angle camera with the 150-degree visual field range, the camera is installed at the position higher than the center of the vehicle body in a downward slightly-deviated 15-degree angle mode, and the camera is connected with an upper computer of the intelligent trolley through a USB interface, so that the information of road conditions such as lane lines, obstacles, traffic lights and the like can be completely collected. The deep stream of Injeida is used as an AI image stream analysis tool, after a real-time road condition image transmitted by a front-view camera is obtained, the image format is properly processed, the image is input into a deep neural network for reasoning, and the type and position size information of all objects in the road condition are obtained after a corresponding post-processing algorithm. The upper computer completes inter-process communication through C++ development, transmits information processed in an AI process to an intelligent trolley control process, and controls the trolley to complete actions such as advancing, parking, turning and the like according to real-time road condition information, so that automatic cruising and intelligent obstacle avoidance of the trolley are completed.
As shown in fig. 1, the embodiment discloses a control method of an automatic cruising intelligent obstacle avoidance trolley, which is characterized in that the intelligent trolley is placed on a driving road of a traffic sand table, and real-time road condition data is acquired through an acquisition unit of the intelligent trolley, wherein the road condition data comprises road condition images; the acquisition unit of intelligent dolly includes the camera. The automatic cruising intelligent obstacle avoidance control of the intelligent trolley comprises the following steps:
s1, analyzing road condition images and identifying information of objects on a road; the information of the object includes a kind, position information, and size information of the object. The object types comprise lane lines and objects with heights, wherein the objects with heights comprise traffic lights, pedestrians and obstacles; the lane lines comprise left straight lines, left turning lines, left right turning lines, right straight lines, right left turning lines and right turning lines; and identifying the lane lines according to the block areas with the preset size. The position information of the object includes center point coordinates of the object; the size information of the object includes a pixel width and a pixel height of the object in the road condition image.
In specific implementation, S1 includes:
s11, clipping the road condition image into the size of INPUT_W multiplied by INPUT_H pixel points, wherein INPUT_W is the width of the road condition image after clipping, and INPUT_H is the height of the road condition image after clipping;
s12, inputting the cut road condition image into a preset deep learning network for object recognition, decoding information under a preset confidence threshold conf_thresh, and obtaining the information of the object in the road condition image through a non-maximum suppression algorithm under a preset NMS threshold nms_thresh.
S2, estimating the distance between the intelligent trolley and each object with the height according to the information of the object identified in the S1 and the pre-stored real height of each object.
The method for estimating the distance dis between the intelligent trolley and the object with the height comprises the following steps:
dis= (true height×f)/height; wherein true height is the real height of the pre-stored object, f is the focal length of the camera of the intelligent trolley, and height is the height of the object detected in S1.
In specific implementation, the measurement process of the focal length f of the camera is shown in fig. 2, and includes:
placing an object, e.g. a cone, at a set given position dis 0 A place; setting a focal length initial value f 0 ；
Obtaining a corresponding distance estimation value (true height×f) according to the height of the object identified in S1 and the recorded true height of the object 0 )/height；
If the estimated value is smaller than dis 0 And the difference value is larger than a preset error value error, and the value f of the focal length is upwardly adjusted to f 0 +△f(△f>0) Recalculating the estimated value; if the estimated value is greater than dis 0 And the difference value is larger than a preset error value error, the value f of the focal length is downwards adjusted to f 0 -△f(△f>0) Recalculating the estimated value; the above procedure is repeated until | (true height×f 0 )/height-dis 0 |≤error。
S3, carrying out driving strategy analysis according to the identified object information and the estimated distance between the intelligent trolley and the object with the height; if the traffic signal lamp in the red light state exists in the identified object and the distance between the traffic signal lamp and the trolley is smaller than the preset red light distance threshold value, or if the pedestrian exists and the distance between the traffic signal lamp and the trolley is smaller than the preset pedestrian distance threshold value, the process goes to S4, and otherwise, the process goes to S5.
S4, executing a preset waiting strategy. In specific implementation, the waiting strategy includes:
if the distance between the trolley and the traffic signal lamp is smaller than a preset red light distance threshold value traffic_thresh, and the current signal lamp type is red light, the trolley is stopped for waiting, and the driving state before the signal lamp is recovered after being converted into a green light;
if the distance between the trolley and the pedestrian is smaller than the preset pedestrian distance threshold ped _thresh, the trolley is stopped for waiting, and the traveling state before the pedestrian is recovered after passing.
S5, if the recognized object has an obstacle and the distance between the recognized object and the trolley is smaller than a preset safe avoidance distance, executing a preset obstacle avoidance driving strategy; otherwise go to S6.
In specific implementation, the process of the obstacle avoidance driving strategy is shown in fig. 3, and includes:
s1, analyzing to obtain position information and size information of the obstacle;
if the coordinates of the central point of the obstacle
Judging that the obstacle is at the left side of the trolley, and controlling the trolley to turn right +.>
An angle, wherein maxTurnAngle represents the maximum steering angle supported by the trolley, and width represents the pixel width of the obstacle in the road condition image; after bypassing the obstacle, controlling the trolley to turn left by the same angle and return to the original running track;
if the coordinates of the central point of the obstacle
Judging that the obstacle is on the right side of the trolley, and controlling the trolley to turn leftwards +.>
An angle, wherein maxTurnAngle represents the maximum steering angle supported by the cart; after the obstacle is bypassed, the trolley is controlled to turn right by the same angle and return to the original running track.
And S6, executing preset autonomous cruise driving.
In specific implementation, the process of the autonomous cruise driving strategy is shown in fig. 4, and includes:
s6.1 divides the lane information identified in S1 into the following lane sets: left straight line: l (L) ls {l 1 ,l 2 ,…,l N1 Left turn line }: l (L) ll {l 1 ,l 2 ,…,l N2 Left-right turn line L lr {l 1 ,l 2 ,…,l N3 Straight line L on right side rs {{l 1 ,l 2 ,…,l N4 }, right left turn line L rl {l 1 ,l 2 ,…,l N5 Right side right turn line L rr {l 1 ,l 2 ,…,l N6 -a }; wherein the elements in each set are the coordinates of the central point of the block-shaped area of the lane line;
s6.2, judging that the current trolley is in a straight-going, left-turning or right-turning state according to the number of block area elements in each lane line set in the step S6.1, and obtaining an X coordinate lane_center X of the center position of the current driving lane. The process of obtaining the X coordinate Lane_center X of the center position of the current driving lane comprises the following steps:
comparing the number of elements in each lane line set, and if the number of elements in the left straight line set is more than that of the left turning line set and the left right turning line set, the number of elements in the right straight line set is more than that of the right left turning line set and the right turning line set; judging that the trolley is in a straight running state, and acquiring a left straight line set L ls Straight line set L on right side rs The method comprises the steps of carrying out a first treatment on the surface of the Will L ls And L rs Fitting to a straight line using the ordinate center Y of the center point as an independent variable and the abscissa center X of the center point as an independent variable X, respectively ls ＝k ls ×centerY ls +b ls ，centerX rs ＝k rs ×centerY rs +b rs The method comprises the steps of carrying out a first treatment on the surface of the Obtaining the coordinates of the position of the headstock at the moment
And the center position X coordinate of the current lane travel +.>
If the left side left turn line set has more elementsA left straight line set and a left right turning line set, wherein the elements in the right left turning line set are more than those in the right straight line set and the right turning line set; judging that the trolley is in a left turning state, and acquiring a left turning line set L ll Set of right left turn lines L rl The method comprises the steps of carrying out a first treatment on the surface of the Will L ll And L rl Fitting to quadratic curves each having a central point ordinate center Y as an independent variable and a central point abscissa center X as a dependent variable X ll ＝k ll1 ×centerY 2 ll +k ll2 ×centerY ll +b ll ，centerX rl ＝k rl1 ×centerY 2 rl +k rl2 ×centerY rl +b rl The method comprises the steps of carrying out a first treatment on the surface of the Obtaining the coordinates of the position of the headstock at the moment as
Center position X coordinate of current lane travel
If the elements in the left-right turning line set are more than the left-left turning line set and the left-right straight line set, the elements in the right-right turning line set are more than the right-left turning line set and the right-right straight line set; judging that the trolley is in a right turning state and acquiring a left-right turning line set L lr Right side right turn line set L rr The method comprises the steps of carrying out a first treatment on the surface of the Will L lr And L rr Fitting to quadratic curves each having a central point ordinate center Y as an independent variable and a central point abscissa center X as a dependent variable X lr ＝k lr1 ×centerY 2 lr +k lr2 ×centerY lr +b lr ，centerX rr ＝k rr1 ×centerY 2 rr +k rr2 ×centerY rr +b rr The method comprises the steps of carrying out a first treatment on the surface of the Obtaining the coordinates of the position of the headstock at the moment as
Center position X coordinate of current lane travel
In S6.2 of the present embodiment, the fitting of the straight line or the quadratic curve is completed by the least square method:
assuming that the current lane line set contains central point coordinate information of N block areas, the order of an equation to be fitted is M, and the following equation is satisfied:
k 1 ×centerY M +k 2 ×centerY M-1 +…+k M+1 ＝centerX；
wherein k is i Coefficients of the secondary terms in the corresponding equation, i e (1, 2 …, m+1);
the equivalent a·w=b is represented by a matrix as follows:
obtaining coefficient matrix W= (A) T A) -1 A T B, and obtaining M+1 coefficients k of M-order equation 1 ，k 2 ，…k M+1 ，
Finally, the equation of M-order curve or straight line is obtained by fitting: centrx=k 1 ×centerY M +k 2 ×centerY M-1 +…+k M+1 。
S6.3 calculating an error value of the center position of the trolley and the current driving lane
And obtaining output quantity turnAngle (t) for controlling steering angle of the steering engine according to the position type PID.
Since the accumulation capacity of the integral term I causes instability of the control system, the PD is directly used for control in this embodiment: turnAngle (t) =K P error(t)+K D [error(t)-error(t-1)]The method comprises the steps of carrying out a first treatment on the surface of the Wherein K is P And K D Respectively a proportional coefficient and a differential coefficient, t representing the moment.
S6.4, obtaining the steering angle which is finally required to be adjusted by the trolley by adopting a dynamic window median filtering mode; the dynamic window median filtering includes: calculating turn angle values at the current n+1 times, taking the median value as an angle value to be adjusted of the trolley, wherein mectinangle=med [ turn angle (t), turn angle (t-1), …, turn angle (t-n) ],0<n; if the value exceeds the maximum steering angle maxTurnAngle supported by the trolley, the adjustment value is constrained to maxTurnAngle;
and S6.5, controlling the trolley to complete corresponding steering actions according to the calculated angle value which is finally required to be adjusted by the trolley in the step S6.4.
By using the method, after the intelligent trolley is placed on a running road of a traffic sand table, real-time road condition images are acquired through the acquisition unit of the intelligent trolley, the road condition images are processed and analyzed, information of objects on the road condition images is extracted, and the distance between the intelligent trolley and the objects with heights is estimated. Therefore, the current road condition of the intelligent trolley can be comprehensively and accurately known, and the corresponding driving strategy is selected. Specifically, if a traffic signal lamp in a red light state exists in the identified object and the distance between the traffic signal lamp and the trolley is smaller than a preset red light distance threshold value, or if a pedestrian exists and the distance between the traffic signal lamp and the trolley is smaller than a preset pedestrian distance threshold value, the intelligent trolley is indicated to have the risk of running the red light or colliding with the pedestrian, so that a predicted waiting strategy is executed, and the running rationality is ensured; if the risk of red light running or collision with pedestrians does not exist, but the recognized object has an obstacle (such as a cone barrel) and the distance between the recognized object and the trolley is smaller than the preset safe avoidance distance, the intelligent trolley can continue to run but needs to avoid the obstacle, so that a preset obstacle avoidance driving strategy is executed, and the running safety is ensured; if there is no risk, it is indicated that the intelligent vehicle can run normally, and thus preset autonomous cruise driving is performed.
Compared with the prior art, the intelligent trolley can complete the cruising task in a more complex semi-open traffic sand table environment, the intelligent trolley can realize functions of intelligent obstacle avoidance, automatic cruising and the like, the limitation that the intelligent trolley can only complete the tracking task in a fixed scene and the obstacle avoidance strategy is single is broken, and the intelligent degree of the intelligent trolley and the strain capacity of traffic events are improved. In addition, the method analyzes and processes the road image in the traffic sand table in real time through a deep learning method and a post-processing algorithm thereof, and compared with the prior art that an infrared obstacle avoidance or ultrasonic obstacle avoidance is used as an obstacle avoidance scheme, the method can more intelligently identify the obstacle to carry out more flexible obstacle avoidance, so that the control mode of the intelligent trolley is more diversified, and the function expansion and secondary development are facilitated; according to the obstacle avoidance driving strategy in the method, the trolley can be guaranteed to fully avoid the obstacle, and the original driving track can be adjusted in time after the obstacle is avoided.
Besides, the method pre-stores the real heights of all objects on the sand table road, and the real distance between the intelligent trolley and the objects can be directly estimated through road condition images. The method also provides a complete autonomous cruise driving strategy, and can ensure that the trolley is positioned at an accurate position as far as possible in the driving process. The autonomous driving strategy is completely based on actual data of each lane line set on the real-time road condition image, has clear logic and simple process, and can ensure the effectiveness and consistency of the autonomous cruise driving process.
Finally, it should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the technical solution, and those skilled in the art should understand that modifications and equivalents may be made to the technical solution of the present invention without departing from the spirit and scope of the present invention, and all such modifications and equivalents are included in the scope of the claims.
Claims (10)
1. The control method of the automatic cruising intelligent obstacle avoidance trolley is characterized in that the intelligent trolley is placed on a driving road of a traffic sand table, real-time road condition data are collected through a collecting unit of the intelligent trolley, and the road condition data comprise road condition images; then, the automatic cruising intelligent obstacle avoidance control of the intelligent trolley comprises the following steps:
s1, analyzing road condition images and identifying information of objects on a road; the information of the object comprises the type of the object, the type of the object comprises a lane line and an object with a height, and the object with the height comprises a traffic signal lamp, pedestrians and obstacles;
s2, estimating the distance between the intelligent trolley and each object with the height according to the information of the object identified in the S1 and the pre-stored real height of each object;
s3, carrying out driving strategy analysis according to the identified object information and the estimated distance between the intelligent trolley and the object with the height; if the traffic signal lamp in the red light state exists in the identified object and the distance between the traffic signal lamp and the trolley is smaller than the preset red light distance threshold value, or if the pedestrian exists and the distance between the traffic signal lamp and the trolley is smaller than the preset pedestrian distance threshold value, the step S4 is carried out, and otherwise, the step S5 is carried out;
s4, executing a preset waiting strategy;
s5, if the recognized object has an obstacle and the distance between the recognized object and the trolley is smaller than a preset safe avoidance distance, executing a preset obstacle avoidance driving strategy; otherwise, turning to S6;
and S6, executing preset autonomous cruise driving.
2. The control method of the automatic cruise intelligent obstacle avoidance trolley according to claim 1, wherein: s1 comprises the following steps:
s11, clipping the road condition image into the size of INPUT_W multiplied by INPUT_H pixel points, wherein INPUT_W is the width of the road condition image after clipping, and INPUT_H is the height of the road condition image after clipping;
s12, inputting the cut road condition image into a preset deep learning network for object recognition, decoding information under a preset confidence threshold conf_thresh, and obtaining the information of the object in the road condition image through a non-maximum suppression algorithm under a preset NMS threshold nms_thresh.
3. The control method of the automatic cruise intelligent obstacle avoidance trolley according to claim 2, wherein: in S1, the information of the object also comprises position information and size information of the object; the position information of the object includes center point coordinates of the object; the size information of the object includes a pixel width and a pixel height of the object in the road condition image.
4. The control method of the automatic cruise intelligent obstacle avoidance vehicle according to claim 3, wherein: in S1, the lane lines include a left straight line, a left turn line, a left right turn line, a right straight line, a right left turn line, and a right turn line; and identifying the lane lines according to the block areas with the preset size.
5. The control method of the automatic cruise intelligent obstacle avoidance trolley according to claim 4, wherein: the acquisition unit of the intelligent trolley comprises a camera; in S2, the estimation method of the distance dis between the intelligent trolley and the object with height is as follows:
dis= (true height×f)/height; wherein true height is the real height of the pre-stored object, f is the focal length of the camera of the intelligent trolley, and height is the height of the object detected in S1.
6. The control method of the automatic cruise intelligent obstacle avoidance trolley according to claim 5, wherein: in S4, the waiting policy includes:
if the distance between the trolley and the traffic signal lamp is smaller than a preset red light distance threshold value traffic_thresh, and the current signal lamp type is red light, the trolley is stopped for waiting, and the driving state before the signal lamp is recovered after being converted into a green light;
if the distance between the trolley and the pedestrian is smaller than the preset pedestrian distance threshold ped _thresh, the trolley is stopped for waiting, and the traveling state before the pedestrian is recovered after passing.
7. The control method of the automatic cruise intelligent obstacle avoidance trolley according to claim 6, wherein: in S5, the obstacle avoidance driving policy includes:
s1, analyzing to obtain position information and size information of the obstacle;
if the coordinates of the central point of the obstacle
Judging that the obstacle is at the left side of the trolley, and controlling the trolley to turn right +.>
An angle, wherein maxTurnAngle represents the maximum steering angle supported by the trolley, and width represents the pixel width of the obstacle in the road condition image; after bypassing the obstacle, controlling the trolley to turn left by the same angle and return to the original running track;
if the coordinates of the central point of the obstacle
Judging that the obstacle is on the right side of the trolley, and controlling the trolley to turn leftwards +.>
An angle, wherein maxTurnAngle represents the maximum steering angle supported by the cart; after the obstacle is bypassed, the trolley is controlled to turn right by the same angle and return to the original running track.
8. The control method of the automatic cruise intelligent obstacle avoidance vehicle according to claim 7, wherein: in S6, the autonomous cruise driving strategy includes:
s6.1 divides the lane information identified in S1 into the following lane sets: left straight line: l (L) ls {l 1 ,l 2 ,…,l N1 Left turn line }: l (L) ll {l 1 ,l 2 ,…,l N2 Left-right turn line L lr {l 1 ,l 2 ,…,l N3 Straight line L on right side rs {{l 1 ,l 2 ,…,l N4 }, right left turn line L rl {l 1 ,l 2 ,…,l N5 Right side right turn line L rr {l 1 ,l 2 ,…,l N6 -a }; wherein the elements in each set are the coordinates of the central point of the block-shaped area of the lane line;
s6.2, judging that the current trolley is in a straight-going, left-turning or right-turning state according to the number of block area elements in each lane line set in the step S6.1, and obtaining an X coordinate lane_center X of the center position of the current driving lane;
s6.3 calculating an error value of the center position of the trolley and the current driving lane
Obtaining output quantity turnAngle (t) for controlling steering angle of the steering engine according to the position type PID; control using PD: turnAngle (t) =K P error(t)+K D [error(t)-error(t-1)]The method comprises the steps of carrying out a first treatment on the surface of the Wherein K is P And K D Respectively a proportional coefficient and a differential coefficient, wherein t represents time;
s6.4, obtaining the steering angle which is finally required to be adjusted by the trolley by adopting a dynamic window median filtering mode; the dynamic window median filtering includes: calculating turn angle values at the current n+1 times, taking the median value as an angle value to be adjusted of the trolley, wherein mectinangle=med [ turn angle (t), turn angle (t-1), …, turn angle (t-n) ],0<n; if the value exceeds the maximum steering angle maxTurnAngle supported by the trolley, the adjustment value is constrained to maxTurnAngle;
and S6.5, controlling the trolley to complete corresponding steering actions according to the calculated angle value which is finally required to be adjusted by the trolley in the step S6.4.
9. The control method of the automatic cruise intelligent obstacle avoidance vehicle according to claim 8, wherein: in S6.2, the process of obtaining the X coordinate lane_center X of the center position of the current driving lane includes:
comparing the number of elements in each lane line set, if the number of elements in the left straight line set is more than the left turning line set and the left right turning line set, and the number of elements in the right straight line set is more than the right left turning line set and the right turning line set, judging that the trolley is in a straight running state, and acquiring a left straight line set L ls Straight line set L on right side rs The method comprises the steps of carrying out a first treatment on the surface of the Will L ls And L rs Fitting to a straight line using the ordinate center Y of the center point as an independent variable and the abscissa center X of the center point as an independent variable X，centerX ls ＝k ls ×centerY ls +b ls ，centerX rs ＝k rs ×centerY rs +b rs The method comprises the steps of carrying out a first treatment on the surface of the Obtaining the coordinates of the position of the headstock at the moment
Input_h), the center position X coordinate of the current lane travel +.>
If the elements in the left turning line set are more than the left straight line set and the left right turning line set, and the elements in the right left turning line set are more than the right straight line set and the right turning line set, the trolley is judged to be in a left turning state, and a left turning line set L is acquired ll Set of right left turn lines L rl The method comprises the steps of carrying out a first treatment on the surface of the Will L ll And L rl Fitting to quadratic curves each having a central point ordinate center Y as an independent variable and a central point abscissa center X as a dependent variable X ll ＝k ll1 ×centerY 2 ll +k ll2 ×centerY ll +b ll ，centerX rl ＝k rl1 ×centerY 2 rl +k rl2 ×centerY rl +b rl The method comprises the steps of carrying out a first treatment on the surface of the Obtaining the coordinates of the position of the headstock at the moment as
Center position X coordinate of current lane travel
If the elements in the left-right turn line set are more than the left-left turn line set and the left-right straight line set, and the elements in the right-right turn line set are more than the right-left turn line set and the right-right straight line set, the trolley is judged to be in the state ofIn the right turning state, a left-right turning line set L is obtained lr Right side right turn line set L rr The method comprises the steps of carrying out a first treatment on the surface of the Will L lr And L rr Fitting to quadratic curves each having a central point ordinate center Y as an independent variable and a central point abscissa center X as a dependent variable X lr ＝k lr1 ×centerY 2 lr +k lr2 ×centerY lr +b lr ，centerX rr ＝k rr1 ×centerY 2 rr +k rr2 ×centerY rr +b rr The method comprises the steps of carrying out a first treatment on the surface of the Obtaining the coordinates of the position of the headstock at the moment as
Center position X coordinate of current lane travel
10. The control method of the automatic cruise intelligent obstacle avoidance vehicle according to claim 9, characterized by: in S6.2, the fitting of the straight line or conic is done by least square method:
assuming that the current lane line set contains central point coordinate information of N block areas, the order of an equation to be fitted is M, and the following equation is satisfied:
k 1 ×centerY M +k 2 ×centerY M-1 +…+k M+1 ＝centerX；
wherein k is i Coefficients of the secondary terms in the corresponding equation, i e (1, 2 …, m+1);
the equivalent a·w=b is represented by a matrix as follows:
obtaining coefficient matrix W= (A) T A) -1 A T B, and obtaining M+1 coefficients k of M-order equation 1 ，k 2 ，…k M+1 Finally fitting to obtain M-orderEquation for curve or straight line: centrx=k 1 ×centerY M +k 2 ×centerY M-1 +…+k M+1 。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310091539.1A CN116360422A (en) | 2023-02-06 | 2023-02-06 | Control method of automatic cruising intelligent obstacle avoidance trolley |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310091539.1A CN116360422A (en) | 2023-02-06 | 2023-02-06 | Control method of automatic cruising intelligent obstacle avoidance trolley |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116360422A true CN116360422A (en) | 2023-06-30 |
Family
ID=86916643
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310091539.1A Pending CN116360422A (en) | 2023-02-06 | 2023-02-06 | Control method of automatic cruising intelligent obstacle avoidance trolley |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116360422A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0789443A (en) * | 1993-07-22 | 1995-04-04 | Nissan Motor Co Ltd | Automatic steering device for vehicle |
| CN105700533A (en) * | 2016-04-22 | 2016-06-22 | 扬州大学 | Agricultural machinery automatic driving control system based on Beidou navigation and method thereof |
| CN206601623U (en) * | 2017-02-28 | 2017-10-31 | 中原工学院 | A kind of big barrier obstruction-avoiding control system of intelligent carriage based on Multi-sensor Fusion |
| CN109782764A (en) * | 2019-01-21 | 2019-05-21 | 湖北汽车工业学院 | A kind of unmanned logistics distribution system of intelligent solar, control method and dispensing vehicle |
| US20200369264A1 (en) * | 2019-05-21 | 2020-11-26 | Mando Corporation | Collision avoidance device, vehicle having the same and method of controlling the vehicle |
| CN112268538A (en) * | 2020-10-20 | 2021-01-26 | 中国人民解放军陆军工程大学 | Geometric optical distance measurement method suitable for small unmanned vehicle |
| CN113570873A (en) * | 2021-07-19 | 2021-10-29 | 东南大学 | Method for detecting distance between front intersection and lamp panel of signal lamp |
-
2023
- 2023-02-06 CN CN202310091539.1A patent/CN116360422A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0789443A (en) * | 1993-07-22 | 1995-04-04 | Nissan Motor Co Ltd | Automatic steering device for vehicle |
| CN105700533A (en) * | 2016-04-22 | 2016-06-22 | 扬州大学 | Agricultural machinery automatic driving control system based on Beidou navigation and method thereof |
| CN206601623U (en) * | 2017-02-28 | 2017-10-31 | 中原工学院 | A kind of big barrier obstruction-avoiding control system of intelligent carriage based on Multi-sensor Fusion |
| CN109782764A (en) * | 2019-01-21 | 2019-05-21 | 湖北汽车工业学院 | A kind of unmanned logistics distribution system of intelligent solar, control method and dispensing vehicle |
| US20200369264A1 (en) * | 2019-05-21 | 2020-11-26 | Mando Corporation | Collision avoidance device, vehicle having the same and method of controlling the vehicle |
| CN112268538A (en) * | 2020-10-20 | 2021-01-26 | 中国人民解放军陆军工程大学 | Geometric optical distance measurement method suitable for small unmanned vehicle |
| CN113570873A (en) * | 2021-07-19 | 2021-10-29 | 东南大学 | Method for detecting distance between front intersection and lamp panel of signal lamp |
Non-Patent Citations (2)
| Title |
|---|
| 江杰，等: "基于多传感器信息融合的惯性导引AGV设计", 自动化应用, no. 10, 31 December 2015 (2015-12-31) * |
| 王铮，等: "基于双目视觉的ＡＧＶ障碍物检测与避障", 计算机集成制造***, vol. 24, no. 2, 28 February 2018 (2018-02-28) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11104327B2 (en) | Method for automated parking of a vehicle | |
| CN113246985B (en) | Intelligent network vehicle merging and changing control method for expressway ramps under mixed-traveling condition | |
| US20200348667A1 (en) | Control system for semi-autonomous control of vehicle along learned route | |
| CN110097785B (en) | Recognition early warning device and early warning method for front vehicle cut-in or emergency lane change | |
| EP3825803B1 (en) | Collision-avoidance system for autonomous-capable vehicles | |
| CN109649393B (en) | Path planning method and device for automatically driving lane change | |
| CN110293970B (en) | Driving control method and device for automatic driving automobile and automobile | |
| JP7128623B2 (en) | Vehicle predictive control | |
| CN106864458B (en) | One kind is automatically around barrier system and method, intelligent automobile | |
| CN106845332B (en) | Vision-based wet road condition detection using tire side splash | |
| CN109386155A (en) | Nobody towards automated parking ground parks the alignment method of transfer robot | |
| RU2719106C1 (en) | Parking control method and parking control equipment | |
| US20210370916A1 (en) | Parking assist apparatus and a method of controlling the parking assist apparatus | |
| CN110796102B (en) | Vehicle target sensing system and method | |
| CN109324616A (en) | Nobody based on onboard sensor parks the alignment method of transfer robot | |
| JP2023512513A (en) | parking assist system | |
| CN112824995A (en) | Method for controlling a motor vehicle and control device | |
| CN113228131A (en) | Method and system for providing ambient data | |
| WO2019166142A1 (en) | Methods and apparatus for acquisition and tracking, object classification and terrain inference | |
| CN107792052B (en) | Someone or unmanned bimodulus steering electric machineshop car | |
| CN113467474A (en) | Automatic driving hierarchical control system | |
| JP2023527599A (en) | Computer-implemented method, non-transitory machine-readable medium, data processing system and computer program for operating an autonomous vehicle | |
| CN116360422A (en) | Control method of automatic cruising intelligent obstacle avoidance trolley | |
| He et al. | Monocular based lane-change on scaled-down autonomous vehicles | |
| CN112351154A (en) | Unmanned automobile road condition recognition system |
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 |