Method, device and equipment for recognizing trailer posture by robot, storage medium and program product

Abstract

The invention relates to a method, device and equipment for recognizing the posture of a trailer through a robot, a storage medium and a program product. The method comprises the following steps: acquiring prior data of a trailer and a radar point cloud acquired by a robot; filtering the radar point cloud based on the detection area of the trailer to obtain a first target radar point cloud; clustering the first target radar point cloud to obtain a first clustering point cloud, and finding out a clustering point pair matched with the prior data in the first clustering point cloud to obtain a first candidate clustering point pair; determining a first angle of the trailer according to the first candidate clustering point pair; and identifying the attitude of the trailer according to the prior data and the first angle. By adopting the method, the robot can accurately and quickly recognize the posture of the trailer.

Description

Technical Field

[0001] This application relates to the field of robot technology, and in particular, to a method, device, equipment, storage medium, and program product for a robot to recognize the attitude of a trailer. Background Art

[0002] With the development of automation and artificial intelligence, more and more robots are applied in production and life. Currently, in many factories, robots are widely used to tow trailers such as material racks or mobile pallets to transport materials or goods. In the existing traction systems, only the perception and obstacle avoidance of the robot are often considered, but the obstacle avoidance problem of the towed trailer is not considered. Summary of the Invention

[0003] Based on this, in view of the above technical problems, it is necessary to provide a method, device, equipment, storage medium, and program product for a robot to recognize the attitude of a trailer, which can enable the robot to accurately and quickly recognize the attitude modeling of the trailer for obstacle avoidance.

[0004] In a first aspect, this application provides a method for a robot to recognize the attitude of a trailer, and the method includes:

[0005] Obtain the prior data of the trailer and the radar point cloud collected by the robot, where the prior data includes the size of the trailer, the preset constraint angle between the trailer and the towing device, and the length of the towing device;

[0006] Filter the radar point cloud based on the detection area of the trailer to obtain a first target radar point cloud;

[0007] Cluster the first target radar point cloud to obtain a first clustered point cloud, and find a clustered point pair that matches the prior data in the first clustered point cloud to obtain a first candidate clustered point pair;

[0008] Determine the first angle of the trailer according to the first candidate clustered point pair;

[0009] Identify the attitude of the trailer according to the prior data and the first angle.

[0010] In a second aspect, this application also provides a device for a robot to recognize the attitude of a trailer, and the device includes:

[0011] An obtaining module, configured to obtain the prior data of the trailer and the radar point cloud collected by the robot, where the prior data includes the size of the trailer, the preset constraint angle between the trailer and the towing device, and the length of the towing device;

[0012] A filtering module, configured to filter the radar point cloud based on the detection area of the trailer to obtain a first target radar point cloud;

[0013] A clustering module, configured to cluster the first target radar point cloud to obtain a first clustered point cloud;

[0014] A searching module, configured to search for clustered point pairs matching the prior data in the first clustered point cloud to obtain a first candidate clustered point pair;

[0015] A determining module, configured to determine a first angle of the trailer according to the first candidate clustered point pair;

[0016] An identifying module, configured to identify the posture of the trailer according to the prior data and the first angle.

[0017] In a third aspect, the present application further provides a computer device, which includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the method for a robot to identify the posture of a trailer are implemented.

[0018] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for a robot to identify the posture of a trailer are implemented.

[0019] In a fifth aspect, the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps of the method for a robot to identify the posture of a trailer are implemented.

[0020] For the above method, device, computer device, storage medium and computer program product for a robot to identify the posture of a trailer, prior data of the trailer and radar point cloud collected by the robot are obtained; the radar point cloud is filtered based on the detection area of the trailer to obtain a first target radar point cloud, thereby reducing the number of radar point clouds, effectively reducing the calculation amount, and facilitating the rapid detection of the posture of the trailer; and, the first target radar point cloud is clustered to obtain a first clustered point cloud, and clustered point pairs matching the prior data are searched in the first clustered point cloud to obtain a first candidate clustered point pair; a first angle of the trailer is determined according to the first candidate clustered point pair; thus, the posture of the trailer can be quickly and accurately identified according to the prior data and the first angle, and then the posture can be used for modeling, which can effectively avoid the trailer hitting an obstacle during travel. Description of the Drawings

[0021] Figure 1 It is a schematic diagram of the application environment of the method for a robot to identify the posture of a trailer in an embodiment;

[0022] Figure 2 It is a schematic flowchart of the method for a robot to identify the posture of a trailer in an embodiment;

[0023] Figure 3 Schematic diagram of a system for a robot to tow a trailer in an embodiment;

[0024] Figure 4 Schematic diagram of a detection area in an embodiment;

[0025] Figure 5 Block diagram of the structure of a device for a robot to recognize the attitude of a trailer in an embodiment;

[0026] Figure 6 Internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0027] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0028] It should be noted that in the description of the present application, the terms "first, second, third and fourth" only distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first, second, third and fourth" can be interchanged with a specific order or sequence when allowed, so that the embodiments of the present application described can be implemented in an order other than the one illustrated or described.

[0029] An embodiment of the present application provides a method for a robot to recognize the attitude of a trailer, which can be applied to an application scenario as shown in Figure 1 In this scenario, the robot 110 towes the trailer 130 through the towing device 120. The robot 110 can be configured as any intelligent device with functions of sensing obstacle avoidance and autonomous navigation, such as an automated guided vehicle (AGV / AMR), a cleaning robot, a delivery robot, a reception robot, etc.; the trailer 130 can be configured as various racks, trolleys, carts, etc. with casters and accommodation spaces. The robot 110 is traction-connected to the trailer 130 through a towing device 120 such as a towing bar, a towing hook, a towing chain, a towing rope, etc. In the illustrated embodiment, a lidar 111 is provided at each of the right front corner and the left rear corner of the robot 110, and the robot 110 obtains radar point cloud data through the lidar 111 for environmental perception.

[0030] As shown in Figure 2 In an embodiment, a method for a robot to recognize the attitude of a trailer is provided. The method can be executed by the robot in Figure 1 and includes the following steps:

[0031] S202. Obtain the prior data of the trailer and the radar point cloud collected by the robot. The prior data includes the size of the trailer, the preset constraint angle between the trailer and the towing device, and the length of the towing device.

[0032] S204. Filter the radar point cloud based on the detection area of the trailer to obtain the first target radar point cloud.

[0033] S206. Cluster the first target radar point cloud to obtain the first clustered point cloud, and search for the clustered point pairs that match the prior data in the first clustered point cloud to obtain the first candidate clustered point pairs. The first clustered point cloud can be multiple clustered clusters (which can be called sub-clustered point clouds) after clustering the first target radar point cloud. The clustered point can be the sub-clustered point cloud in the first clustered point cloud. Correspondingly, the clustered point pair can be a combination of two sub-clustered point clouds in the first clustered point cloud.

[0034] S208. Determine the first angle of the trailer according to the first candidate clustered point pair. The first angle can be the angle of the trailer relative to the robot, such as the angle of the trailer relative to the robot.

[0035] S210. Identify the posture of the trailer according to the prior data and the first angle.

[0036] In the above embodiments, the prior data of the trailer and the radar point cloud collected by the robot are obtained; the radar point cloud is filtered based on the detection area of the trailer to obtain the first target radar point cloud, thereby reducing the number of radar point clouds and effectively reducing the computational amount, which is beneficial to quickly detecting the posture of the trailer; in addition, the first target radar point cloud is clustered to obtain the first clustered point cloud, and the clustered point pairs that match the prior data are searched in the first clustered point cloud to obtain the first candidate clustered point pairs; the first angle of the trailer is determined according to the first candidate clustered point pairs; thus, the posture of the trailer can be quickly and accurately identified according to the prior data and the first angle, and then the posture can be used for modeling, which can effectively avoid the trailer hitting obstacles during driving.

[0037] As Figure 3 shown, the size of the trailer is used to characterize the size of the trailer, which includes the length (L) and width (W) of the trailer; the preset constraint angle (θ) between the trailer and the towing device is used to characterize the deviation angle of the trailer relative to the robot, that is, the included angle between the trailer and the towing device. This included angle can be preset to 80° to 100° or 85° to 95°, and the specific range can be adjusted according to the actual constraint requirements.

[0038] In one embodiment, in response to a data input operation, the robot acquires prior data for the trailer that is input. For example, the robot displays an input page on the screen, and the user can input at least one of the dimensions of the trailer, a preset constraint angle between the trailer and the towing device, or the length of the towing device on the input page.

[0039] In one embodiment, the robot emits a laser beam through the installed lidar. When the laser beam encounters an "obstacle", it is reflected, so that the reflected laser beam can be collected to obtain radar point clouds. Among them, the obstacle can include the trailer, and can also include other objects that collide with the laser beam, such as environmental objects and pedestrians.

[0040] Preferably, reflective plates or reflective stickers can be provided on the two side edges of the front side of the trailer, so that the robot can identify the distance between the two reflective plates or reflective stickers through the radar device to identify the width of the trailer, and then combine the length data of the trailer to determine the position of the trailer.

[0041] For example, during the process of the trailer being towed by the robot and moving forward, the trailer may swing left and right within a certain angle range. Using the dimensions of the trailer and the preset constraint angle, the area where the reflective plate may be located (i.e., the detection area) can be obtained, such as Figure 4 the C-shaped area; in addition, the detection area where the trailer may be located can also be obtained using the dimensions of the trailer and the preset constraint angle.

[0042] In one embodiment, the first target radar point cloud in S204 can be the radar point cloud located within the detection area obtained by filtering out the radar point cloud located outside the detection area.

[0043] In one embodiment, the detection area can be determined according to the dimensions of the trailer and the preset constraint angle.

[0044] It can be understood that during the process of the trailer being towed by the robot and moving forward, due to obstacle avoidance actions such as the robot turning or braking, a certain deviation will occur between the trailer and the robot; or, due to uneven ground, the trailer may deviate left and right from directly behind the robot, resulting in a certain angle between the trailer and the robot. Therefore, the trailer is not always directly behind the robot, but may appear directly behind or diagonally behind the robot. Therefore, in order to reduce the calculation amount and ensure the realization of the attitude detection of the trailer, a detection area can be determined behind the robot (including directly behind and diagonally behind), and this detection area is a C-shaped area behind the robot, which can be used to represent the set of positions where the trailer may appear, so as to use the radar point cloud located within the detection area for attitude detection.

[0045] In one embodiment, after obtaining the first target radar point cloud in S204, radar point clouds in the first target radar point cloud with reflection intensities not meeting the reflection intensity threshold may also be filtered out to obtain a second target radar point cloud; clustering is performed on the second target radar point cloud to obtain a second clustered point cloud; and a pair of clustered points matching the prior data is searched for from the second clustered point cloud.

[0046] Further, if no pair of clustered points matching the prior data is found from the second clustered point cloud, S206 is continued to be executed. It can be understood that when the robot collects radar point cloud data through a lidar, due to too low reflection intensities or incorrect information corresponding to certain angles, it may result in no pair of clustered points matching the prior data being found in the second target radar point cloud after intensity filtering and clustering.

[0047] Among them, the first clustered point cloud in S206 includes a "point type clustered point cloud" and a "line type clustered point cloud". In one embodiment, specifically, S206 may include:

[0048] Pairwise combination is performed on each clustered point in the point type clustered point cloud and each edge clustered point in the line type clustered point cloud to obtain a plurality of pairs of clustered points;

[0049] The third angle between the connection line of each pair of clustered points in the plurality of pairs of clustered points and the virtual traction device is determined, where the virtual traction device is the virtual connection line between the midpoint of the connection line and the traction point (m) of the robot, that is, the connection line between the midpoint of the connection line of the pair of clustered points and the traction point (m) of the robot.

[0050] According to the matching degree between the third angle and the preset constraint angle, the matching degree between the length of the virtual traction device and the length of the traction device, the matching degree between the length of the connection line and the width of the trailer, and the matching degree between the first angle and the first angle calculated in the previous frame, a first candidate pair of clustered points is selected from the first clustered point cloud.

[0051] Among them, the edge clustered points in the line type clustered point cloud include a front end point (i.e., a front end clustered point) and a rear end point (i.e., a rear end clustered point). When combining the clustered points, pairwise combination can be performed on each clustered point in the point type clustered point cloud, on the clustered point in the point type clustered point cloud and the front end point, on the clustered point in the point type clustered point cloud and the rear end point, and on the front end point and the rear end point in the point type clustered point cloud.

[0052] For example, if 20 pairs of clustered points are obtained by combining the clustered points in the first clustered point cloud, then pairs of clustered points satisfying the following conditions are selected as candidate pairs of clustered points from these 20 pairs of clustered points:

[0053] a) Whether the angle between the virtual traction device between the midpoint of the connection line and the robot traction point and the connection line matches the preset constraint angle;

[0054] b) Whether the length of the virtual traction device between the midpoint of the connection line and the robot traction point is close to the length of the traction device in the prior data;

[0055] c) Whether the length of the connection line is close to the width of the trailer in the prior data;

[0056] d) Whether the angle between the virtual traction device and the connection line is close to the angle detected in the previous frame.

[0057] In one embodiment, after selecting the candidate cluster point pairs and before determining the first angle of the trailer according to the first candidate cluster point pair, it further includes further screening based on the scores obtained by combining the above conditions. Specifically: The robot determines the fifth score according to the matching degree between the third angle and the preset constraint angle, determines the sixth score according to the matching degree between the length of the virtual traction device and the length of the traction device, determines the seventh score according to the matching degree between the length of the connection line and the width of the trailer, and determines the eighth score according to the matching degree between the third angle and the third angle calculated in the previous frame; perform a weighted sum of the fifth score, the sixth score, the seventh score, and the eighth score to obtain the second comprehensive score; screen the first candidate cluster point cloud based on the second comprehensive score to obtain the screened first candidate cluster point pair. For example, take the first candidate cluster point pair with the largest second comprehensive score as the screened first candidate cluster point pair, update the screened first candidate cluster point pair to the first candidate cluster point, and then determine the first angle of the trailer according to the updated first candidate cluster point pair.

[0058] For the divided point type cluster point cloud and line type cluster point cloud, it can be judged in the following way: The robot calculates the circumscribed rectangle for each sub-cluster point cloud in the first cluster point cloud; determine the sub-cluster point cloud with the long side of the circumscribed rectangle less than the preset threshold as the "point type cluster point cloud"; determine the sub-cluster point cloud with the long side of the circumscribed rectangle greater than or equal to the preset threshold as the "line type cluster point cloud".

[0059] In one embodiment, on the contrary, if a cluster point pair matching the prior data can be found from the second cluster point cloud, then take the matching cluster point pair as the second candidate cluster point pair; determine the second angle of the trailer according to the second candidate cluster point pair; identify the attitude of the trailer according to the prior data and the second angle.

[0060] In one embodiment, after selecting the candidate cluster point pairs and before determining the second angle of the trailer based on the second candidate cluster point pairs, the robot further includes further screening based on the scores obtained by combining the above conditions. Specifically, the robot determines a first score according to the matching degree between the first angle and the preset constraint angle, determines a second score according to the matching degree between the length of the virtual towing device and the length of the towing device, determines a third score according to the matching degree between the length of the connection line and the width in the dimensions, and determines a fourth score according to the matching degree between the first angle and the first angle calculated in the previous frame; performs a weighted sum of the first score, the second score, the third score, and the fourth score to obtain a first comprehensive score; screens the second candidate cluster point pairs based on the first comprehensive score to obtain the screened second candidate cluster points. For example, the second candidate cluster point pair with the largest first comprehensive score in the second candidate cluster point pairs is used as the screened second candidate cluster point pair, and the screened second candidate cluster point pair is updated to the second candidate cluster point, and then the second angle of the trailer is determined according to the updated second candidate cluster point pair.

[0061] In the above embodiment, after filtering the radar point cloud based on the detection area to obtain the first target radar point cloud, the first target radar point cloud is further filtered according to the reflection intensity, so as to filter out the point cloud that does not meet the intensity condition, thereby accelerating the speed of attitude recognition; in addition, the obtained second target radar point cloud is used for clustering to obtain a second cluster point cloud, and a matching cluster point pair is selected from the second cluster point cloud as the second candidate cluster point pair, and the second angle of the trailer is determined according to the second candidate cluster point pair; thus, the attitude of the trailer can be quickly and accurately recognized according to the prior data and the second angle, and then the attitude is used for modeling, which can effectively avoid the trailer hitting an obstacle during the driving process.

[0062] Combined with a specific example, the technical solution of the present application is described as follows:

[0063] S1, input the length (L) and width (W) of the trailer, the preset constraint angle (θ) between the trailer and the towing device, and the length (D) of the towing device;

[0064] S2, calculate a mask according to the preset constraint angle and the length (L) and width (W) of the trailer, and select the radar point cloud on the mask;

[0065] Among them, the above mask refers to the detection area, which is a set of all possible positions of the trailer when considering different angles of the trailer.

[0066] S3, filter out the radar point cloud outside the mask to obtain the first cluster point cloud, that is, the radar point cloud located within the mask, denoted as mask_filter_points;

[0067] S4. Filter out the radar point clouds in mask_filter_points with a radar reflection intensity below 235 to obtain the second clustered point cloud, that is, the radar point clouds with a radar reflection intensity above 235, denoted as inte_filter_points;

[0068] S5. First, cluster inte_filter_points to obtain a set of clustered point clouds Φ;

[0069] S6. Judge any two clustered points in the radar clustered point clouds within the set of clustered point clouds Φ to find out the candidate clustered point pairs φ. Among them, the conditions for judgment are as follows:

[0070] S61) Whether the angle between the virtual towing device between the midpoint of the connection line and the robot towing point and this connection line matches the preset constraint angle θ;

[0071] S62) Whether the length of the virtual towing device between the midpoint of the connection line and the robot towing point is close to the length of the towing device in the prior data;

[0072] S63) Whether the length of the connection line is close to the width of the trailer in the prior data;

[0073] S64) Whether the angle between the virtual towing device and this connection line is close to the angle detected in the previous frame.

[0074] S7. Perform weighted scoring on the condition matching degree of the previous step in the candidate clustered point pairs φ, and select the clustered point pair with the highest score as the best candidate clustered point pair;

[0075] S8. If no candidate clustered point pairs that meet the conditions are found in inte_filter_points, then use mask_filter_points for searching;

[0076] S9. Cluster mask_filter_points to obtain a set of clustered point clouds Ω;

[0077] S10. Calculate the minimum bounding rectangle for each radar clustered point cloud within the set of clustered point clouds Ω, and judge the point type clustered point clouds and line type clustered point clouds according to the side lengths of the minimum bounding rectangle. Specifically as follows:

[0078] 1) When the long side of the minimum bounding rectangle is greater than the preset threshold, it is considered that this clustered point cloud is a line type clustered point cloud. Find the two edge points of this cluster according to the angles of each point in the line type clustered point cloud, and record them as the "front end point" and "rear end point" according to the angles of the edge points;

[0079] 2) When the long side of the minimum bounding rectangle is less than the preset threshold, it is considered that this clustered point cloud is a point type clustered point cloud.

[0080] S11. Pairwise combine the rules for point - type clustering point cloud - point - type clustering point cloud, front - end point - rear - end point, point - type clustering point cloud - rear - end point, and front - end point - point - type clustering point cloud, and use the following conditions to find candidate clustering point pairs ω:

[0081] S111) Whether the angle between the virtual traction device between the mid - point of the connection line and the robot traction point and the connection line matches the preset constraint angle θ;

[0082] S112) Whether the length of the virtual traction device between the mid - point of the connection line and the robot traction point is close to the length D of the traction device in the prior data;

[0083] S113) Whether the length of the connection line is close to the width W of the trailer in the prior data;

[0084] S114) Whether the angle between the virtual traction device and the connection line is close to the angle detected in the previous frame.

[0085] S12. Weight the condition matching degrees of the candidate clustering point pairs and select the one with the highest score as the best candidate clustering point pair;

[0086] S13. Obtain the angle of the trailer according to the best candidate clustering point, and calculate the overall attitude of the trailer based on the size and angle of the trailer.

[0087] It should be understood that although the steps in the flowcharts involved in the above - mentioned embodiments are shown in sequence according to the arrows, these steps do not necessarily need to be executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps does not have a strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above - mentioned embodiments may include multiple steps or multiple stages. These steps or stages do not necessarily need to be executed at the same time, but can be executed at different times. The execution order of these steps or stages does not necessarily need to be sequential, but can be executed alternately or in turn with at least a part of other steps or steps in other steps.

[0088] Based on the same inventive concept, an embodiment of the present application further provides a device for a robot to recognize the attitude of a trailer for implementing the method for a robot to recognize the attitude of a trailer involved above. The solution provided by this device to solve the problem is similar to the solution described in the above - mentioned method. Therefore, the specific limitations in one or more embodiments of the device for a robot to recognize the attitude of a trailer provided below can refer to the limitations on the method for a robot to recognize the attitude of a trailer in the above text, and will not be repeated here.

[0089] In one embodiment, as Figure 5As shown in the figure, a device for a robot to recognize the attitude of a trailer is provided, including: an acquisition module 502, a filtering module 504, a clustering module 506, a searching module 508, a determination module 510, and an identification module 512, where:

[0090] The acquisition module 502 is configured to acquire the prior data of the trailer and the radar point cloud collected by the robot. The prior data includes the size of the trailer, the preset constraint angle between the trailer and the towing device, and the length of the towing device.

[0091] The filtering module 504 is configured to filter the radar point cloud based on the detection area of the trailer to obtain the first target radar point cloud.

[0092] The clustering module 506 is configured to cluster the first target radar point cloud to obtain the first clustered point cloud.

[0093] The searching module 508 is configured to search for the clustered point pairs matching the prior data in the first clustered point cloud to obtain the first candidate clustered point pairs.

[0094] The determination module 510 is configured to determine the first angle of the trailer according to the first candidate clustered point pairs.

[0095] The identification module 512 is configured to identify the attitude of the trailer according to the prior data and the first angle.

[0096] In one embodiment, the filtering module is further configured to filter out the radar point cloud located outside the detection area according to the detection area of the trailer determined by the size and preset constraint angle of the trailer, so as to obtain the first target radar point cloud.

[0097] In one embodiment, the device further includes:

[0098] The filtering-out module is configured to filter out the radar point cloud with a reflection intensity not meeting the reflection intensity threshold in the first target radar point cloud to obtain the second target radar point cloud.

[0099] The clustering module is further configured to cluster the second target radar point cloud to obtain the second clustered point cloud.

[0100] The searching module is further configured to search for the clustered point pairs matching the prior data from the second clustered point cloud.

[0101] The clustering module is further configured to, if no clustered point pairs matching the prior data are found from the second clustered point cloud, continue to cluster the first target radar point cloud to obtain the first clustered point cloud.

[0102] The searching module is further configured to search for the clustered point pairs matching the prior data in the first clustered point cloud to obtain the first candidate clustered point pairs.

[0103] In one embodiment, the device further includes:

[0104] A determination module, further configured to determine a first angle between the connection line of each pair of clustering points in the second clustering point cloud and the virtual towing device, where the virtual towing device is a virtual connection line between the midpoint of the connection line and the towing point of the robot;

[0105] A search module, further configured to search for a pair of clustering points that matches the prior data from the second clustering point cloud based on the matching degree between the first angle and a preset constraint angle, the matching degree between the length of the virtual towing device and the length of the towing device, the matching degree between the length of the connection line and the width in the dimensions of the trailer, and the matching degree between the first angle and the first angle calculated in the previous frame;

[0106] The determination module, further configured to: if a pair of clustering points that matches the prior data is found from the second clustering point cloud, use the matching pair of clustering points as the second candidate pair of clustering points; determine a second angle of the trailer according to the second candidate pair of clustering points;

[0107] An identification module, further configured to identify the attitude of the trailer according to the prior data and the second angle.

[0108] In one embodiment, the search module is further configured to determine a first score according to the matching degree between the first angle and the preset constraint angle, determine a second score according to the matching degree between the length of the virtual towing device and the length of the towing device, determine a third score according to the matching degree between the length of the connection line and the width in the dimensions, and determine a fourth score according to the matching degree between the first angle and the first angle calculated in the previous frame; perform a weighted sum of the first score, the second score, the third score, and the fourth score to obtain a first comprehensive score; and screen the second candidate pair of clustering points based on the first comprehensive score to obtain the screened second candidate clustering points.

[0109] In the above embodiment, after filtering the radar point cloud based on the detection area to obtain the first target radar point cloud, the first target radar point cloud is further filtered according to the reflection intensity, so as to filter out the point cloud that does not meet the conditions, thereby accelerating the speed of attitude recognition; and, clustering the obtained second target radar point cloud to obtain a second clustering point cloud, selecting a matching pair of clustering points from the second clustering point cloud as the second candidate pair of clustering points, and determining a second angle of the trailer according to the second candidate pair of clustering points; thus, the attitude of the trailer can be quickly and accurately recognized according to the prior data and the second angle, and then modeling can be performed using the attitude, which can effectively avoid the trailer hitting an obstacle during the traveling process.

[0110] In one embodiment, the first clustering point cloud includes a point type clustering point cloud and a line type clustering point cloud;

[0111] The searching module is further configured to: pair each clustering point in the point-type clustering point cloud with each edge clustering point in the line-type clustering point cloud pairwise to obtain a plurality of clustering point pairs; determine a third angle between the connection line of each clustering point pair in the plurality of clustering point pairs and the virtual traction device, where the virtual traction device is a virtual connection line between the midpoint of the connection line and the traction point of the robot; select a first candidate clustering point pair from the first clustering point cloud according to the matching degree between the third angle and a preset constraint angle, the matching degree between the length of the virtual traction device and the length of the traction device, the matching degree between the length of the connection line and the width in the dimension, and the matching degree between the first angle and the first angle calculated in the previous frame.

[0112] In one embodiment, the searching module is further configured to: determine a fifth score according to the matching degree between the third angle and the preset constraint angle, determine a sixth score according to the matching degree between the length of the virtual traction device and the length of the traction device, determine a seventh score according to the matching degree between the length of the connection line and the width in the dimension, and determine an eighth score according to the matching degree between the third angle and the third angle calculated in the previous frame; perform weighted summation on the fifth score, the sixth score, the seventh score, and the eighth score to obtain a second comprehensive score; and screen the first candidate clustering point cloud based on the second comprehensive score to obtain the screened first candidate clustering point pair.

[0113] In one embodiment, the device further includes:

[0114] A calculating module, configured to calculate an external circumscribed rectangle for each sub-clustering point cloud in the first clustering point cloud;

[0115] A determining module, configured to determine the sub-clustering point cloud with the long side of the external circumscribed rectangle less than a preset threshold as the point-type clustering point cloud, and determine the sub-clustering point cloud with the long side of the external circumscribed rectangle greater than or equal to the preset threshold as the line-type clustering point cloud.

[0116] Each module in the above device for a robot to recognize the attitude of a trailer can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in the form of hardware or be independent of the processor, or be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above respective modules.

[0117] In one embodiment, a computer device is provided. The computer device can be a robot, and its internal structure diagram can be as Figure 6As shown in the figure. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface, the display unit, and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFL (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for a robot to recognize the attitude of a trailer. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0118] Those skilled in the art can understand that Figure 6 the structure shown in the figure is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0119] In one embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, it implements the steps of the above method for a robot to recognize the attitude of a trailer.

[0120] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, it implements the steps of the above method for a robot to recognize the attitude of a trailer.

[0121] In one embodiment, a computer program product is provided, including a computer program. When the computer program is executed by the processor, it implements the steps of the above method for a robot to recognize the attitude of a trailer.

[0122] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., and are not limited thereto. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., and are not limited thereto.

[0123] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0124] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for a robot to identify a trailer posture, characterized in that: The method comprises: Acquire prior data of the trailer and a radar point cloud collected by the robot, wherein the prior data includes the size of the trailer, a preset constraint angle between the trailer and the traction device, and the length of the traction device; Filtering the radar point cloud based on the detection area of ​​the trailer to obtain a first target radar point cloud; Clustering the first target radar point cloud to obtain a first cluster point cloud, and searching for cluster point pairs matching the priori data in the first cluster point cloud to obtain first candidate cluster point pairs; determining a first angle of the trailer according to the first candidate cluster point pair; The posture of the trailer is identified according to the priori data and the first angle.

2. The method according to claim 1, characterized in that The filtering the radar point cloud based on the detection area of ​​the trailer to obtain the first target radar point cloud includes: determining a detection area of ​​the trailer according to the size and the preset constraint angle; The radar point cloud outside the detection area is filtered out to obtain the first target radar point cloud.

3. The method according to claim 1 or 2, characterized in that: The method further comprises: Filtering out radar point clouds whose reflection intensities do not meet a reflection intensity threshold in the first target radar point cloud to obtain a second target radar point cloud; Clustering the second target radar point cloud to obtain a second cluster point cloud, and searching for cluster point pairs matching the prior data from the second cluster point cloud; If no cluster point pair matching the prior data is found in the second cluster point cloud, continue to perform the step of clustering the first target radar point cloud to obtain a first cluster point cloud, and find a cluster point pair matching the prior data in the first cluster point cloud to obtain a first candidate cluster point pair.

4. The method according to claim 3, characterized in that: The step of searching the second cluster point cloud for a cluster point pair matching the priori data comprises: Determine a first angle between a line connecting each pair of cluster points in the second cluster point cloud and a virtual traction device, wherein the virtual traction device is a virtual connecting line between a midpoint of the line and a traction point of the robot; Based on the matching degree between the first angle and the preset constraint angle, the matching degree between the length of the virtual traction device and the length of the traction device, the matching degree between the length of the connecting line and the width in the dimension, and the matching degree between the first angle and the first angle calculated in the previous frame, cluster point pairs matching the prior data are searched from the second cluster point cloud.

5. The method according to claim 3, characterized in that: The method further comprises: If a cluster point pair matching the priori data is found in the second cluster point cloud, the matching cluster point pair is used as a second candidate cluster point pair; determining a second angle of the trailer according to the second candidate cluster point pair; The posture of the trailer is identified according to the priori data and the second angle.

6. The method according to claim 5, characterized in that Before determining the second angle of the trailer according to the second candidate cluster point pair, the method further includes: Determine a first score according to the matching degree between the first angle and the preset constraint angle, determine a second score according to the matching degree between the length of the virtual traction device and the length of the traction device, determine a third score according to the matching degree between the length of the connecting line and the width in the dimension, and determine a fourth score according to the matching degree between the first angle and the first angle calculated in the previous frame; Performing a weighted summation on the first score, the second score, the third score, and the fourth score to obtain a first comprehensive score; The second candidate cluster point pairs are screened based on the first comprehensive score to obtain screened second candidate cluster points.

7. The method according to claim 1 or 2, characterized in that: The first cluster point cloud includes a point type cluster point cloud and a line type cluster point cloud; the step of searching for a cluster point pair matching the priori data in the first cluster point cloud to obtain a first candidate cluster point pair includes: Combining each cluster point in the point type cluster point cloud with each edge cluster point in the line type cluster point cloud in pairs to obtain a plurality of cluster point pairs; Determine a third angle between a line connecting each of the plurality of cluster point pairs and a virtual traction device, wherein the virtual traction device is a virtual connecting line between a midpoint of the line and a traction point of the robot; A first candidate cluster point pair is selected from the first cluster point cloud based on the matching degree between the third angle and the preset constraint angle, the matching degree between the length of the virtual traction device and the length of the traction device, the matching degree between the length of the connecting line and the width in the dimension, and the matching degree between the first angle and the first angle calculated in the previous frame.

8. The method according to claim 7, characterized in that Before determining the first angle of the trailer according to the first candidate cluster point pair, the method further includes: A fifth score is determined according to a matching degree between the third angle and the preset constraint angle, a sixth score is determined according to a matching degree between the length of the virtual traction device and the length of the traction device, a seventh score is determined according to a matching degree between the length of the connecting line and the width in the dimension, and an eighth score is determined according to a matching degree between the third angle and the third angle calculated in the previous frame; Performing a weighted summation on the fifth score, the sixth score, the seventh score, and the eighth score to obtain a second comprehensive score; The first candidate cluster point cloud is screened based on the second comprehensive score to obtain screened first candidate cluster point pairs.

9. The method according to claim 7, characterized in that: The method further comprises: Calculating a bounding rectangle for each sub-cluster point cloud in the first cluster point cloud; Determine the sub-cluster point cloud whose long side of the circumscribed rectangle is smaller than a preset threshold as the point type cluster point cloud; The sub-cluster point cloud whose long side of the circumscribed rectangle is greater than or equal to the preset threshold is determined to be the line type cluster point cloud.

10. A robot device for identifying trailer posture, characterized in that: The device comprises: An acquisition module, used to acquire prior data of the trailer and a radar point cloud collected by the robot, wherein the prior data includes the size of the trailer, a preset constraint angle between the trailer and the traction device, and the length of the traction device; A filtering module, configured to filter the radar point cloud based on the detection area of ​​the trailer to obtain a first target radar point cloud; A clustering module, used for clustering the first target radar point cloud to obtain a first clustered point cloud; A search module, used to search for cluster point pairs matching the priori data in the first cluster point cloud to obtain first candidate cluster point pairs; a determination module, configured to determine a first angle of the trailer according to the first candidate cluster point pair; A recognition module is used to recognize the posture of the trailer according to the prior data and the first angle.

11. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

13. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

Images