AGV following method based on radar inter-frame difference maximum line area

Abstract

The application discloses an AGV following method based on radar interframe difference maximum line area, which comprises the following steps: controlling a laser radar to scan and obtain a previous frame of point cloud data array and store the previous frame of point cloud data array in a point cloud data buffer area, and continuously scanning to obtain a current frame of point cloud data array; performing difference processing on the current frame of point cloud data array and the previous frame of point cloud data array; calculating and judging the maximum line area of continuous '1' in the difference result as a calibration following target; determining the polar radius and polar angle of the target following point and comparing the polar radius and polar angle with preset target following distance threshold and polar angle threshold to realize target following; after confirming that an AGV reaches the target following position, clearing the point cloud data buffer area, and obtaining a frame of complete radar point cloud data as the previous frame of point cloud data for next calibration following target identification; and the application reduces the requirements for the operation speed of a processor and the memory when the AGV follows, and improves the real-time performance, stability and accuracy of following when the target moves fast and is large in size.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of AGV, robot, industrial robot arm, intelligent control of autonomous vehicle, in particular to an AGV following method based on radar inter-frame difference maximum line area. BACKGROUND

[0002] In the manufacturing industry, logistics and other industries, AGV trolley based on radar following method plays an important role in the handling link, solving the problems of large labor demand, low efficiency and easy repetition in handling.

[0003] However, at present, there are some problems in the application of radar following technology in real scene. In the scene of multiple obstacles, most of the existing radar target recognition methods extract the feature data of different targets by clustering method, construct a feature database, then classify in the classifier, and finally identify the specific target, so it is necessary to establish a feature database and the algorithm is complex, which requires high operation speed and memory of the processor, so the real-time performance of following is weak; it is easy to lose the target when identifying high-speed moving objects, and the accuracy of identifying high-speed moving objects is low.

[0004] Therefore, it is necessary to develop a following method to solve the above technical problems, reduce the requirements of AGV trolley on processor operation speed and memory, and improve the real-time performance, stability and accuracy when following large targets in a multi-obstacle scene. SUMMARY

[0005] The present application aims to overcome the shortcomings of the prior art, and provides an AGV following method based on radar inter-frame difference maximum line area, which reduces the requirements of AGV trolley on processor operation speed and memory and improves the real-time performance, stability and accuracy when AGV trolley follows large targets in a multi-obstacle scene.

[0006] The technical scheme adopted by the present application is as follows:

[0007] An AGV following method based on radar inter-frame difference maximum line area, comprising the steps of:

[0008] S1, controlling the laser radar to continuously scan to obtain environment point cloud data, the environment point cloud data comprising the polar angle and distance information of the obstacle in the radar polar coordinate system, storing the scanned previous frame point cloud data array into the point cloud data buffer area, and the laser radar continues to scan until the complete current frame point cloud data array is obtained.

[0009] S2, comparing the current frame point cloud data array with the previous frame point cloud data array performing comparison to get a difference discrimination bit array performing comparison to get a difference discrimination bit array ;

[0010] S3, analyzing and calculating the difference discrimination bit array the maximum line area of continuous "1" judging the maximum line area of continuous "1" corresponding to the calibration following target;

[0011] S4, taking the center scanning point of the calibration following target as a target following point performing comparison between the polar distance and the polar angle of the target following point and the preset target following distance threshold and the polar angle threshold to realize target following;

[0012] S5, after confirming that the AGV reaches the target following position, clearing the point cloud data buffer area, and obtaining a complete frame of radar point cloud data as the previous frame of point cloud data for the next calibration following target identification.

[0013] Further, in the S1 step, the laser radar is a single-line laser radar; the origin of the radar polar coordinate system is the origin of the AGV coordinate system, the polar axis is the longitudinal coordinate axis of the AGV coordinate system, the positive direction of the polar angle is counterclockwise, and the scanning distance of the laser radar to the obstacle is the polar radius, and the polar radius is inf when the obstacle exceeds the radar measurement range; the memory size of the point cloud data buffer area is the size of a frame of point cloud data.

[0014] Further, in the S2 step, performing difference between the current frame of point cloud data array and the previous frame of point cloud data array and performing comparison with a target determination threshold to get a difference discrimination bit array , specifically including steps:

[0015] The difference determination algorithm formula is:

[0016] ;

[0017] Wherein is the data in the address of the current frame of point cloud data array , is the data in the address of the previous frame of point cloud data array , is the target determination threshold;

[0018] like , in the differential discriminant bit array The address t is assigned a value of 1. , preliminarily judged as the target following point;

[0019] like , in the differential discriminant bit array The address t is assigned a value of 0. , it is judged that it is not a target following point;

[0020] Finally, the differential discriminant bit array is obtained .

[0021] Furthermore, the target determination threshold Multiply the desired calibration target speed by the coefficient And the time difference between the current frame and the previous frame .

[0022] Furthermore, in step S3, the differential discriminant bit array is analyzed and calculated. The maximum linear area of ​​consecutive "1"s in , determine the maximum linear area of ​​the continuous "1" The corresponding point cloud is the calibration target, which includes the following steps:

[0023] (1) Using loop traversal and pointer method, the differential discriminant bit array According to the principle of continuous identical elements, group and sum them, and compare the size of each group in turn to preliminarily determine the maximum linear area of ​​the continuous "1" for the calibration following target;

[0024] (2) Setting the area threshold for the target line to be followed To prevent misidentification, if , then determine the maximum linear area of ​​the continuous "1" The corresponding point cloud is the calibration follow target; if , it is determined to be a misidentified target, that is, the calibrated follow target is not identified this time and the next calibrated follow target identification is performed again.

[0025] Furthermore, the differential discrimination bit array is arrayed by using a loop traversal and pointer method. The method groups and sums consecutive identical elements, and compares the sizes of each group in turn. The specific steps include:

[0026] (1) Set the pointer variable to point to the differential discriminant bit array ;

[0027] (2) Determine the bit array from the difference The first address starts to loop and judge and accumulate. If the address is judged Elements within and the latter address Elements within If they are equal and equal to 1, then accumulate the elements and continue to judge until an element is 0 and stop accumulating to get the area of ​​the previous continuous 1 line ;

[0028] (3) Continue to judge until the element is 1, and then repeat the cycle judgment accumulation to obtain the current continuous 1 line area and the area of ​​the previous continuous 1 line Compare the size, retain the larger value and assign it to , while retaining the address of the first element of the group corresponding to the larger value and assigning it to ;

[0029] (4) Continue the loop comparison until the loop ends, and finally get the maximum line area of ​​the continuous "1" , and the maximum linear area of ​​the continuous "1" The address of the first element in the corresponding group is .

[0030] Furthermore, in step S4, the center scanning point of the calibrated follow target is used as the target follow point. , the target follows the point Polar diameter and polar angle The preset target following distance threshold and polar angle threshold Comparison is performed to achieve target following, specifically including the following steps:

[0031] (1) Calculate the center scanning point from the calibrated follow target as the target follow point , determine the target following point The polar diameter in the radar polar coordinate system and polar angle :

[0032] Calculate the target following point Address , get the current frame point cloud data array The address Data within and the previous frame point cloud data array The address Data within ;

[0033] like and , then the target follows the point Polar diameter Pick and The smaller value of ;

[0034] like and , then the target follows the point Polar diameter ;

[0035] like and , then the target follows the point Polar diameter ;

[0036] The target following point Polar angle Follow the target point Address Multiply by the radar angular resolution;

[0037] (2) Setting the target following distance threshold and the polar angle threshold and follow the point with the target Polar diameter and polar angle Compare to realize the motion control of AGV car:

[0038] like and , the system determines that the AGV needs to be controlled to rotate counterclockwise or turn left ;

[0039] like and , the system determines that the AGV needs to be controlled to rotate clockwise or turn right ;

[0040] like and , the system determines that the AGV needs to be controlled to rotate clockwise or turn right ;

[0041] like and , the system determines that the AGV needs to be controlled to rotate counterclockwise or turn left ;

[0042] like , the system determines that the AGV needs to move closer to the target following point , control the AGV car to translate towards the direction of the target following point ; ;

[0043] If , the system determines that the AGV car needs to move away from the target following point , control the AGV car to translate in the opposite direction of the target following point . .

[0044] Further, in the S5 step, the AGV car reaching the target following position is confirmed, specifically comprising the steps of:

[0045] The position of the current radar polar coordinate system relative to the initial radar polar coordinate system is estimated by the AGV car positioning method, and coordinate transformation is performed to obtain the real-time transformation coordinates of the current radar polar coordinate system relative to the initial radar polar coordinate system , and the target transformation coordinates are continuously compared, and the AGV car continues to move until and , it is determined that the AGV car has reached the target following position.

[0046] Further, the AGV following method based on the maximum line area of the radar inter-frame difference provided by the application is not only suitable for target following of the AGV car, but also suitable for target following of robots, industrial robot arms and autonomous driving cars.

[0047] The application has the following beneficial effects:

[0048] (1) The AGV following method based on the maximum line area of the radar inter-frame difference provided by the application obtains the maximum line area of the continuous "1" as the calibration following target by a series of processes such as difference, threshold comparison and loop iteration on the adjacent two frames of point cloud data, the application mainly processes the bit array, and only needs to set the memory as a point cloud data buffer area with the size of one frame of point cloud data, therefore, the algorithm of the application is simple to realize, the required memory is small, the requirements of the AGV car on the operation speed of the processor and the memory are reduced, the real-time following of the AGV car is strengthened, and the cost is reduced;

[0049] (2) When a faster and larger target is identified, the maximum line area of the continuous "1" will also be larger, therefore, in a multi-obstacle scene, the faster and larger target can be accurately identified and is not easy to lose;

[0050] (3) The application sets the polar angle threshold The AGV trolley can realize 360-degree omnidirectional posture following, and the AGV trolley is convenient for loading and unloading, and the application is suitable for target following of the AGV trolley, and can also be suitable for target following of robots, industrial mechanical arms and self-driving cars.

[0051] The application will be further described in detail below according to the drawings and embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to make the technical solutions of the embodiments of the application more clearly understood, the drawings needed to be used in the related technical description of the embodiments will be briefly introduced below, and obviously, the drawings in the following description can be used to obtain other drawings without creative labor for those skilled in the art.

[0053] Figure 1 A flowchart of an AGV following method based on radar interframe difference maximum line area provided by the embodiment of the application is shown in the figure.

[0054] Figure 2 A structural schematic diagram of a Mecanum wheel type AGV trolley provided by the embodiment of the application is shown in the figure.

[0055] Figure 3 A driving integrated module structural schematic diagram of a Mecanum wheel type AGV trolley provided by the embodiment of the application is shown in the figure.

[0056] Figure 4 A radar interframe difference determination schematic diagram of a Mecanum wheel type AGV trolley provided by the embodiment of the application is shown in the figure.

[0057] Figure 5 A target identification and post-rotation control schematic diagram of a Mecanum wheel type AGV trolley provided by the embodiment of the application is shown in the figure.

[0058] In the figure: 1, aluminum profile frame; 2, STM32 control box; 3, driving integrated module; 4, inertial sensor; 5, industrial computer; 6, motor driver; 7, ultrasonic module; 8, laser radar; 9, servo motor; 10, speed reducer; 11, L-shaped connecting plate; 12, shaft coupling; 13, bearing seat; 14, Mecanum wheel; 15, module mounting plate. DETAILED DESCRIPTION

[0059] In order to make the technical solutions of the embodiments of the application more clearly understood, the drawings needed to be used in the related technical description of the embodiments will be briefly introduced below, and obviously, the drawings in the following description can be used to obtain other drawings without creative labor for those skilled in the art.

[0060] Reference Figure 2 and Figure 3The embodiment of the present application provides a Mecanum wheel type AGV trolley, which comprises a trolley body, a laser radar 8 arranged at the top center of the trolley body, a control system arranged in the trolley body and connected with the laser radar 8, and an auxiliary sensing system arranged in and around the trolley body.

[0061] The trolley body comprises an aluminum profile frame 1 and a driving integrated module 3.

[0062] The aluminum profile frame is connected and fastened by aluminum profile materials, aluminum corner connectors and T-shaped bolts.

[0063] The driving integrated module comprises a servo motor 9, a speed reducer 10, an L-shaped connecting plate 11, a shaft coupling 12, a bearing seat 13, a Mecanum wheel 14 and a module mounting plate 15.

[0064] The laser radar 8 is an N10 TOF single-line laser radar, the radar scanning frequency is 10HZ, the radar measurement range radius is 0.5-11m, and the radar angle resolution is 0.5°.

[0065] The control system is used for target identification, controls the Mecanum wheel type AGV trolley to follow the target movement and stops when encountering an obstacle in self-following.

[0066] The control system comprises an STM32 control box 2, an industrial computer 5 and a motor driver 6.

[0067] The chip in the STM32 control box 2 is an STM32F103ZET6.

[0068] The industrial computer 5 runs a ROS system.

[0069] The auxiliary sensing system comprises an inertial sensor 4 and an ultrasonic module 7.

[0070] Referring to Figure 1 、 Figure 2 、 Figure 3 、 Figure 4 and Figure 5 , the AGV following method based on radar interframe difference maximum line area provided by the present application is implemented in the Mecanum wheel type AGV trolley provided by the present embodiment, and comprises the following steps:

[0071] S101: control the laser radar 8 to continuously scan to obtain environment point cloud data, the environment point cloud data comprising polar angle and distance information of an obstacle in a radar polar coordinate system, store a previous frame point cloud data array scanned into a point cloud data buffer area, and the laser radar 8 continues to scan until a complete current frame point cloud data array is obtained. ;

[0072] S102: the current frame point cloud data array​ with the previous frame point cloud data array and a target determination threshold value for comparison, obtaining a difference discrimination bit array ;

[0073] S103: analyze and calculate the difference discrimination bit array The maximum line area of continuous "1" , judge the maximum line area of continuous "1" The corresponding point cloud is the calibration following target;

[0074] S104: taking the center scanning point of the calibration following target as the target following point , the polar radius and the polar angle of the target following point are compared with the preset target following distance threshold value and the polar angle threshold value to realize target following;

[0075] S105: after confirming that the Mecanum wheel type AGV reaches the target following position, clear the point cloud data buffer area, and obtain a frame of complete radar point cloud data as the previous frame point cloud data for the next calibration following target identification.

[0076] In this embodiment, the laser radar 8 is installed at the top center of the vehicle body of the Mecanum wheel type AGV, so that the origin and polar axis x of the radar polar coordinate system coincide with the origin and longitudinal coordinate axis Y of the Mecanum wheel type AGV coordinate system, the positive direction of the polar angle of the radar polar coordinate system is counterclockwise, and the scanning distance of the laser radar to the obstacle is the polar radius. When the obstacle exceeds the measurement range of the radar, the polar radius is inf.

[0077] In this embodiment, the industrial computer 5 controls the laser radar 8 to continuously scan to obtain environmental point cloud data. In order to reduce the memory requirement, the point cloud data buffer area only needs to be set to the memory size of a frame of point cloud data.

[0078] In this embodiment, the industrial computer 5 differentiates the current frame point cloud data array with the previous frame point cloud data array and a target determination threshold value for comparison, obtaining a difference discrimination bit array , which specifically includes the following steps:

[0079] The difference determination algorithm formula is:

[0080] ;

[0081] in The point cloud data array of the current frame At the address The data within The point cloud data array of the previous frame At the address The data within is the target determination threshold;

[0082] like , in the differential discriminant bit array The address t is assigned a value of 1. , preliminarily judged as the target following point;

[0083] like , in the differential discriminant bit array The address t is assigned a value of 0. , it is judged that it is not a target following point;

[0084] Finally, the differential discriminant bit array is obtained .

[0085] In this embodiment, the industrial computer 5 analyzes and calculates the differential discriminant bit array The maximum linear area of ​​consecutive "1"s in , determine the maximum linear area of ​​the continuous "1" The corresponding point cloud is the calibration target, which includes the following steps:

[0086] (1) Using loop traversal and pointer method, the differential discriminant bit array According to the principle of continuous identical elements, group and sum them, and compare the size of each group in turn to preliminarily determine the maximum linear area of ​​the continuous "1" for the calibration following target;

[0087] (2) Setting the area threshold for the target line to be followed To prevent misidentification, if , then determine the maximum linear area of ​​the continuous "1" The corresponding point cloud is the calibration follow target; if , it is determined to be a misidentified target, that is, the calibrated follow target is not identified this time and the next calibrated follow target identification is performed again.

[0088] In this example, the differential discrimination bit array is arrayed using a loop traversal and pointer method. The method groups and sums consecutive identical elements, and compares the sizes of each group in turn. The specific steps include:

[0089] (1) Set a pointer variable to point to the difference discrimination bit array ;

[0090] (2) Start from the first address of the difference discrimination bit array , and perform a loop judgment and accumulation. If the elements in the judgment address and the elements in the next address are equal to 1, the elements are accumulated, and the judgment continues until the elements are 0, and the previous continuous 1 line area is obtained ; ;

[0091] (3) Continue to judge until the element is 1, and then re-perform the loop judgment and accumulation to obtain the current continuous 1 line area , and compare it with the previous continuous 1 line area . The larger value is retained and assigned to , and the address of the first element of the group corresponding to the larger value is retained and assigned to ;

[0092] (4) Continue to compare in a loop until the loop ends, and finally obtain the maximum line area of the continuous "1" , and the address of the first element in the group corresponding to the maximum line area of the continuous "1" is .

[0093] As shown in Figure 4 , in this embodiment, the specific principle of identifying the expected calibration following target through radar inter-frame difference discrimination includes:

[0094] The target judgment threshold is the speed of the expected calibration following target multiplied by a coefficient and the time difference between the current frame and the previous frame , that is , wherein the coefficient , according to the radar scanning frequency of 10HZ, the time difference ;

[0095] In the previous frame, the laser radar 8 scans the and on the expected calibration following target, and the distance is inf , in the current frame, the laser radar 8 scans the and on the expected calibration following target, and the distance is inf ​, where inf means the distance of the scanning point is infinite;

[0096] because and The distance is inf, so that the desired calibration follows the non-overlapping area before and after the target movement and The difference between the two frames before and after and Greater than the target judgment threshold , similarly the non-overlapping area and The difference distance between all the previous and next frames is greater than the target judgment threshold , and because the coefficient , you can set the appropriate coefficient Ensure that the desired calibration follows the overlapping area before and after the target movement The difference distance between all the previous and next frames includes are greater than the target judgment threshold Finally, according to the differential determination algorithm formula, the non-overlapping area and and the overlapping area All of them are judged as "1", so the faster the speed and the larger the volume of the desired calibration tracking target, the larger the area of ​​the continuous "1" line of the desired calibration tracking target will be.

[0097] In this embodiment, the industrial computer 5 selects the central scanning point of the calibrated follow target as the target follow point. , calculate the target following point Polar diameter and polar angle , specifically including the steps:

[0098] Calculate the target following point Address , get the current frame point cloud data array The address Data within and the previous frame point cloud data array The address Data within ;

[0099] like and , then the target follows the point Polar diameter Pick and The smaller value of ;

[0100] like and , then the target follows the point Polar diameter ;

[0101] like and , then the target follows the point Polar diameter ;

[0102] The target following point Polar angle Follow the target point Address Multiply by the radar angular resolution ;

[0103] Wherein, inf is the polar diameter when the obstacle exceeds the radar measurement range, and the radar angular resolution is 0.5°.

[0104] like Figure 5 As shown, in this embodiment, the target following distance threshold is set and the polar angle threshold and follow the point with the target Polar diameter and polar angle Compare and realize the rotation and translation control of the Mecanum wheeled AGV:

[0105] like and , the system determines that the Mecanum wheeled AGV needs to be controlled to rotate counterclockwise , with the polar angle threshold and target follow point For example, and , then the Mecanum wheeled AGV needs to rotate counterclockwise ;

[0106] like and , the system determines that the Mecanum wheeled AGV needs to be controlled to rotate clockwise , with the polar angle threshold and target follow point For example, and , then the Mecanum wheeled AGV needs to rotate clockwise ;

[0107] like and , the system determines that the Mecanum wheeled AGV needs to be controlled to rotate clockwise , with the polar angle threshold and target follow point For example, and , then the Mecanum wheeled AGV needs to rotate clockwise ;

[0108] like and , the system determines that the Mecanum wheeled AGV needs to be controlled to rotate counterclockwise , with the polar angle threshold and target follow point For example, and , then the Mecanum wheeled AGV needs to rotate counterclockwise ;

[0109] like , the system determines that the Mecanum wheeled AGV needs to move closer to the target following point , control the Mecanum wheeled AGV to follow the target point Direction translation ;

[0110] like , the system determines that the Mecanum wheeled AGV needs to move away from the target following point , control the Mecanum wheeled AGV to follow the target point The opposite direction of translation .

[0111] In this embodiment, the rotation control is to keep the target following the point The polar angle in the polar coordinate system of the Mecanum wheeled AGV radar is the polar angle threshold The translation control can be decomposed into the control of the horizontal axis and the vertical axis Y axis in the Mecanum wheeled AGV trolley coordinate system, the purpose of which is to keep the target following point The distance from the origin of the polar coordinate system of the Mecanum wheeled AGV radar is the target following distance threshold .

[0112] In this embodiment, the target following point is selected After that, the Mecanum wheeled AGV needs to reach the target following position. The specific steps include:

[0113] (1) According to the target following point Polar angle and the target transformation coordinates The industrial computer 5 calculates the lateral velocity value of the Mecanum wheeled AGV in the initial Mecanum wheeled AGV coordinate system , the longitudinal velocity value of the longitudinal coordinate axis Y axis , the angular velocity value around the vertical coordinate axis Wherein, n is the time required for the Mecanum wheeled AGV to reach the target following position.

[0114] (2) The industrial computer 5 and the STM32 control box 2 establish communication with the STM32F103ZET6, and the STM32F103ZET6 receives the lateral velocity value , the longitudinal velocity value And the angular velocity value Data information sent by the industrial computer 5;

[0115] (3) The STM32F103ZET6 solves the inverse kinematics equation of the Mecanum wheeled AGV, obtains the speed information required by the servo motor 9 and sends it to the motor driver 6, so as to realize the motion control of the Mecanum wheeled AGV;

[0116] (4) The industrial computer 5 determines that the Mecanum wheeled AGV has reached the target following position:

[0117] The current radar polar coordinate system of the Mecanum wheeled AGV relative to the initial radar polar coordinate system is estimated by the fusion positioning algorithm of IMU and wheeled odometry, and coordinate transformation is performed to obtain the real-time transformation coordinates of the current radar polar coordinate system relative to the initial radar polar coordinate system The information of the IMU is measured by the inertial sensor 4, and the information of the wheeled odometry is fed back by the motor driver 6. The industrial computer 5 adopts PID control strategy, continuously compares the real-time transformation coordinates With the target transformation coordinates , so as to continue to control the motion of the Mecanum wheeled AGV until And It is determined that the Mecanum wheeled AGV has reached the target following position.

[0118] In the embodiment, if the Mecanum wheeled AGV reaches the target following position, the point cloud data buffer is cleared, and a frame of complete radar point cloud data is obtained as the previous frame of point cloud data for the next target identification in the calibration following; if the Mecanum wheeled AGV encounters an obstacle before reaching the target following position, the straight-line distance between the Mecanum wheeled AGV and the surrounding obstacle is detected by the ultrasonic module 7, and when the straight-line distance is less than the set safety distance, the Mecanum wheeled AGV is controlled to stop to ensure safety.

[0119] In summary, the embodiment of the present application provides a Mecanum wheeled AGV for realizing an AGV following method based on radar inter-frame difference maximum line area, which can accurately follow a target with high speed and large volume in a multi-obstacle scene and can realize 360° omnidirectional attitude following, and the target is not easy to be lost in the following process, and the real-time performance of the following is strong and the safety is high.

[0120] The above only describes specific embodiments of the present disclosure and is not intended to limit the present disclosure, and any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. An AGV following method based on radar inter-frame difference maximum line area, characterized in that, The method comprises the steps of: S1, control the laser radar to carry out continuous scanning and obtain environment point cloud data, the environment point cloud data includes the polar angle and distance information of the obstacle in the radar polar coordinate system, the scanned previous frame point cloud data array is stored in the point cloud data buffer area, and the laser radar continues to scan until the complete current frame point cloud data array is obtained ; S2, the current frame point cloud data array The point cloud data array of the previous frame Difference and target determination threshold Compare and get the difference discriminant bit array ; S3, analyze and calculate the difference discrimination bit array the maximum line area of continuous "1" , determine the maximum line area of continuous "1" The point cloud corresponding to the calibration following target; S4, taking the central scanning point of the calibrated follow target as the target follow point , the target follows the point Polar diameter and polar angle The preset target following distance threshold and polar angle threshold Make comparisons to achieve goal following; S5, after confirming that the AGV reaches the target following position, clearing the point cloud data buffer area and obtaining a frame of complete radar point cloud data as the previous frame of point cloud data for the next target identification.

2. The AGV following method based on radar inter-frame difference maximum line area according to claim 1, characterized in that, In the S1 step, the laser radar is a single-line laser radar; the origin of the radar polar coordinate system is the origin of the AGV coordinate system, the polar axis is the longitudinal coordinate axis of the AGV coordinate system, the positive direction of the polar angle is the counterclockwise direction, and the scanning distance of the laser radar to the obstacle is the polar radius, and the polar radius is inf when the obstacle exceeds the radar measurement range; the memory size of the point cloud data buffer area is the size of a frame of point cloud data.

3. The AGV following method based on radar inter-frame difference maximum line area according to claim 1, characterized in that, In the S2 step, the current frame point cloud data array is differentiated with the previous frame point cloud data array and compared with a target determination threshold to obtain a differentiated determination bit array , specifically comprising the steps of: The difference determination algorithm formula is: ； wherein is the current frame point cloud data array is the data in the address , is the previous frame point cloud data array is the data in the address , is the target decision threshold; If , the address t in the difference discrimination bit array is assigned 1, that is , the target following point is preliminarily judged; If , the address t in the difference discrimination bit array is assigned 0, that is , it is determined that it is not a target following point. The final difference discrimination bit array is obtained .

4. The AGV following method based on radar inter-frame difference maximum line area according to claim 3, characterized in that, the target determination threshold a speed at which the target is desired to be calibrated to follow multiplied by a coefficient and a time difference between the current frame and the previous frame .

5. The radar frame difference maximum line area based AGV following method according to claim 1, wherein, In the S3 step, the difference discrimination bit array is analyzed and calculated Maximum line area of continuous "1" The maximum line area of the continuous "1" is determined The point cloud corresponding to the maximum line area of the continuous "1" is determined as the calibration following target, specifically comprising the steps of: (1) using the way of loop and pointer, the difference discrimination bit array Grouping sum according to the principle of continuous same elements, and comparing the size of each group in turn, preliminarily determine the maximum line area of the continuous "1" For the calibration of the target; (2) Set the calibration follow target line area threshold For preventing misrecognition, if The maximum line area of the continuous "1" is determined The corresponding point cloud is the calibration follow target; if It is judged as a misrecognition target, that is, the calibration follow target is not recognized this time and the next calibration follow target recognition is re-performed.

6. The AGV following method based on radar inter-frame difference maximum line area according to claim 5, characterized in that, The difference discrimination bit array is traversed in a loop and by means of a pointer The grouping summation is performed according to the principle of continuous same elements, and size comparison is performed on each group in turn, specifically comprising the steps of: (1) setting a pointer variable to point to the difference discrimination bit array ; (2) from the start of the first address of the difference discrimination bit array Loop judgment accumulation, if the element In the judgment address And the element In the next address Equal to 1, then the element accumulation, continue to judge until there is an element for 0 stop accumulation, get the previous continuous 1 line area ; (3) continue to judge until the element is 1, re-perform the loop judgment accumulation, and obtain the current continuous 1 line area And the previous continuous 1 line area Compare the size, keep the larger value and assign it to At the same time, keep the address of the first element of the larger value corresponding group and assign it to ; (4) continue to compare in a loop until the end of the loop, and finally obtain the maximum line area of the continuous "1" , and the address of the first element in the group corresponding to the maximum line area of the continuous "1" is .

7. The radar frame difference maximum line area based AGV following method according to claim 1, wherein, In the S4 step, taking the center scanning point of the calibration target following target as a target following point , the polar radius and the polar angle of the target following point are compared with the preset target following distance threshold value and the polar angle threshold value to achieve target following, specifically including the steps of: (1) calculating a center scan point as a target following point from the calibration following target , determining a polar radius and a polar angle of the target following point in the radar polar coordinate system and :​ Calculate the target following point Address , get the current frame point cloud data array The address Data within and the previous frame point cloud data array The address Data within ; like and , then the target follows the point Polar diameter Pick and The smaller value of ; If and , then the polar radius of the target following point is ; If and then the polar radius of the target following point is ; the target following point polar angle the target following point address multiplied by radar angular resolution; (2) setting the target follow distance threshold and the polar angle threshold , and comparing the polar radius and polar angle of the target follow point to achieve AGV motion control: If and , the system determines that the AGV needs to rotate counterclockwise or turn left ; If and , the system determines that the AGV needs to rotate clockwise or turn right ; If and , the system determines that the AGV needs to rotate clockwise or turn right ; If and , the system determines that the AGV needs to rotate counterclockwise or turn left ; If , the system determines that the AGV needs to approach the target following point , controls the AGV to translate in the direction of the target following point ; ; If , the system determines that the AGV needs to move away from the target following point , controls the AGV to translate in the opposite direction of the target following point . .

8. The AGV following method based on radar inter-frame difference maximum line area according to claim 1, characterized in that, In the S5 step, the AGV reaching the target following position is confirmed, specifically comprising the steps of: The position of the current radar polar coordinate system relative to the initial radar polar coordinate system is estimated by an AGV positioning method, and coordinate transformation is performed to obtain real-time transformation coordinates of the current radar polar coordinate system relative to the initial radar polar coordinate system , and the target transformation coordinates are compared continuously, and the AGV vehicle motion is continued until and , it is determined that the AGV vehicle has reached the target following position.

9. The AGV following method based on radar inter-frame difference maximum line area according to claim 1, characterized in that, Not only suitable for target following of AGV, but also suitable for target following of robots, industrial robot arms and self-driving cars.

Images