Train bottom inspection robot bottom inspection control method and device, robot and storage medium

Abstract

Embodiments of the present application provide a train bottom inspection robot bottom inspection control method, device, robot and storage medium, relating to the technical field of robot control. The method first sets a preset parking point and a preset collection and photographing point of the bottom inspection robot; when the bottom inspection robot reaches the preset parking point, three-dimensional point cloud information of the current bogie of the train bottom is acquired, and the three-dimensional point cloud information is registered with teaching three-dimensional point cloud information through a point cloud registration algorithm to obtain a deviation RT of the bottom inspection robot between the preset parking point and a teaching parking point; the preset collection and photographing point is corrected, path planning is performed according to the corrected collection and photographing point to generate an optimal planning path; and bottom inspection is performed according to the corrected collection and photographing point and the optimal planning path. Through the present application, the deviation is compensated, the finally collected data and the standard data are highly matched, and the success rate of the train bottom inspection algorithm is greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot control, in particular to a train bottom inspection robot bottom inspection control method and device, a robot and a storage medium. BACKGROUND

[0002] Currently, the inspection of the bottom of the train is generally performed by manual inspection, rail-mounted train bottom inspection robot inspection or non-rail-mounted train bottom inspection robot inspection. Among them, the bottom inspection robot is used to detect the train in the tunnel at the bottom of the train.

[0003] For the rail-mounted train bottom inspection robot, due to the different parking positions of the train each time, the bottom inspection robot needs to be manually moved to the head position to execute the pre-taught work file. However, since the position of the head is determined manually, there is an error in the arrival of the bottom inspection robot at the specified working position. If the mechanical arm reaches the specified position according to the manually taught point, the three-dimensional image of the collection point and the collection point at the time of teaching is inconsistent, which will cause the subsequent algorithm to fail to match the standard image during the inspection, resulting in the loss of collected image and three-dimensional point cloud information, and causing the algorithm to fail. In addition, the mechanical arm must run the trajectory according to the manual teaching, which requires manual judgment of the optimal path of the mechanical arm during the manual teaching process. It cannot guarantee the optimal motion path and motion time of the mechanical arm. Moreover, the mechanical arm must collect all the taught points, and cannot selectively collect some points.

[0004] For the non-rail-mounted train bottom inspection robot, there are still the following problems: since the navigation algorithm relies on external devices such as IMU odometry and laser radar, the final parking position still has a certain deviation due to the influence of the calibration error of the positioning device, the environment, the uneven road surface of the tunnel, the calculation error of the car chassis itself, etc. It cannot guarantee the precise arrival at the parking position.

[0005] For the above problem of not being able to accurately arrive at the parking position, since there is an error between the parking position and the parking position at the time of teaching, if the mechanical arm is not used for correction, the collected image will deviate from the standard image. SUMMARY

[0006] In order to solve one of the above technical defects, the present application provides a train bottom inspection robot bottom inspection control method, device, robot and storage medium.

[0007] According to a first aspect of the present application, a train bottom inspection robot bottom inspection control method is provided, which comprises:

[0008] Set the preset parking point of the bottom inspection robot and the preset collection and photographing point of the mechanical arm of the bottom inspection robot under the preset parking point;

[0009] When the bottom inspection robot reaches the preset parking point, the actual three-dimensional point cloud information of the current bogie of the train is obtained through the fine scanning camera, the actual three-dimensional point cloud information is registered with the pre-collected teaching three-dimensional point cloud information through a point cloud registration algorithm, and the deviation RT of the bottom inspection robot between the preset parking point and the teaching parking point is obtained.

[0010] The preset collection and photographing point is corrected according to the deviation RT to obtain a corrected collection and photographing point, and path planning is performed according to the corrected collection and photographing point to generate an optimal planning path without collision.

[0011] The bottom of the train is inspected according to the corrected collection and photographing point and the optimal planning path.

[0012] In an optional embodiment of the present application, the step of registering the actual three-dimensional point cloud information with the pre-collected teaching three-dimensional point cloud information through a point cloud registration algorithm to obtain the deviation RT of the bottom inspection robot between the preset parking point and the teaching parking point in the method further includes:

[0013] The deviation RT is a 4X4 coordinate transformation matrix.

[0014] In an optional embodiment of the present application, the step of correcting the preset collection and photographing point according to the deviation RT to obtain the corrected collection and photographing point in the method further includes:

[0015] The RT coordinate transformation based on the coordinate system of the mechanical arm is performed on all preset collection and photographing points according to the deviation RT.

[0016] In an optional embodiment of the present application, the step of performing path planning according to the corrected collection and photographing point to generate an optimal planning path without collision in the method further includes:

[0017] In the planning space of the mechanical arm, the position of the non-obstacle area is determined, a random point is generated, and growth is performed.

[0018] The growth includes:

[0019] The existing nodes on the growth tree are traversed with the random point as the target, the distance from each existing node to the random point is calculated, and the nearest point is selected.

[0020] The line connecting the nearest point and the random point is taken as the growth direction, the growth point is generated from the nearest point to the target point, and it is determined whether the growth point collides with the obstacle.

[0021] If no collision occurs, the growth point is added to the growth tree.

[0022] If a collision occurs, the growth point is removed, a new random point is generated, and growth is performed according to the new random point to obtain an optimal planning path.

[0023] In an optional embodiment of the present application, the growing in the method further includes: the step size of the growing is a fixed value.

[0024] In an optional embodiment of the present application, the step of registering the actual three-dimensional point cloud information and the pre-acquired teaching three-dimensional point cloud information by the point cloud registration algorithm in the method further includes:

[0025] The corresponding relationship of the points on the two point clouds of the actual three-dimensional point cloud information and the teaching three-dimensional point cloud information is found in a greedy manner, and the rotation relationship parameter R and the translation relationship parameter t are further calculated, and the point cloud registration is performed according to the rotation relationship parameter R and the translation relationship parameter t.

[0026] In an optional embodiment of the present application, the method further includes:

[0027] When the preset parking point of the bottom inspection robot is set, the position of the chassis of the bottom inspection robot relative to the preset starting point of the bottom inspection robot is recorded to set the preset parking point.

[0028] According to a second aspect of the embodiments of the present application, a train bottom inspection robot bottom inspection control device is provided, which includes a preset point setting module, a deviation RT calculation module, a deviation correction path planning module and a bottom inspection module; wherein,

[0029] The preset point setting module is configured to set the preset parking point of the bottom inspection robot and the preset collection and photographing point of the mechanical arm of the bottom inspection robot under the preset parking point.

[0030] The deviation RT calculation module is configured to, when the bottom inspection robot reaches the preset parking point, acquire actual three-dimensional point cloud information of the current train bogie by the precision scanning camera, register the actual three-dimensional point cloud information and pre-acquired teaching three-dimensional point cloud information by a point cloud registration algorithm, and obtain the deviation RT of the bottom inspection robot between the preset parking point and the teaching parking point.

[0031] The deviation correction path planning module is configured to correct the preset collection and photographing point according to the deviation RT to obtain a corrected deviation collection and photographing point, plan a path according to the corrected deviation collection and photographing point, and generate a collision-free optimal planning path.

[0032] The bottom inspection module is configured to perform bottom inspection on the train according to the corrected deviation collection and photographing point and the optimal planning path.

[0033] According to a third aspect of the embodiments of the present application, a train bottom inspection robot is provided, comprising: a memory; a processor; and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement the steps of the method of any one of the first aspect of the embodiments of the present application.

[0034] According to a fourth aspect of the embodiments of the present application, a computer readable storage medium is provided, and a computer program is stored on the computer readable storage medium; the computer program is executed by a processor to implement the steps of the method of any one of the first aspect of the embodiments of the present application.

[0035] The train bottom inspection robot bottom inspection control method provided in the embodiments of the present application has the following beneficial effects:

[0036] The method of the present application is proposed for the problem of error in the parking position of the track type bottom inspection robot and the non-track type bottom inspection robot each time. When the parking position of the robot chassis occurs error, the present application performs registration through the point cloud information of the initial collection position and the standard collection position, obtains the deviation of the robot chassis, and makes a compensation to the deviation through the mechanical arm, so that the finally collected data and the standard data are highly matched, and the success rate of the train bottom inspection algorithm is greatly improved. BRIEF DESCRIPTION OF DRAWINGS

[0037] The accompanying drawings, which are included to provide a further understanding of the present application, constitute a part of the present application, and the illustrative embodiments of the present application and their descriptions serve to explain the present application and do not constitute improper limitations on the present application. In the drawings:

[0038] Figure 1 The flowchart of the train bottom inspection robot bottom inspection control method provided for the embodiments of the present application;

[0039] Figure 2 The coordinate transformation schematic diagram provided for the embodiments of the present application;

[0040] Figure 3 The path planning schematic provided for the embodiments of the present application Figure 1 ;

[0041] Figure 4 The path planning schematic provided for the embodiments of the present application Figure 2 ;

[0042] Figure 5 The path planning schematic provided for the embodiments of the present application Figure 3 ;

[0043] Figure 6 The path planning schematic provided for the embodiments of the present application Figure 4 ;

[0044] Figure 7 A path planning schematic provided for an embodiment of the present application Figure 5

[0045] Figure 8 A structural diagram of a train bottom inspection robot bottom inspection control device provided for an embodiment of the present application

[0046] Figure 9 An internal structure schematic diagram of a bottom inspection robot provided for an embodiment of the present application DETAILED DESCRIPTION

[0047] In order to make the technical solutions and advantages in the embodiments of the present application clearer, the exemplary embodiments of the present application are further described in detail below with reference to the drawings. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0048] Currently, the train bottom is generally inspected by manual inspection, track type train bottom inspection robot inspection or non-track type train bottom inspection robot inspection.

[0049] The track type train bottom inspection robot inspection refers to laying a track with a magnetic strip on the bottom of the train, and moving the robot chassis to the designated position. The robot is manually controlled to reach the train bottom inspection position, the distance from the current chassis position to the train head is recorded through the bottom magnetic strip, and then the movement trajectory of the mechanical arm is recorded to generate a work file through the mechanical arm teaching mode. When the train arrives at the inspection point again, since the position of the train on the track is different each time, the bottom inspection robot is first manually moved to the head of the train, and then moved to the designated position through the pre-recorded inspection stop position. The mechanical arm executes the pre-taught movement trajectory work file, collects the points that need to be photographed, saves the 2D picture and 3D point cloud information of the corresponding points, and transmits them to the algorithm for train unqualified inspection.

[0050] ​The non-orbital train bottom inspection robot is positioned and navigated in real time by a distance sensor to reach a specified parking point. A human controls the robot to reach the train bottom inspection position, records the distance from the current chassis to the train head through the bottom magnetic strip, and then records the motion trajectory of the mechanical arm to generate a work file through the mechanical arm teaching method. When the train reaches the inspection point again, the bottom inspection robot is first moved to the head of the train by a human, and then moved to the specified position by the pre-recorded inspection parking point position. The mechanical arm executes the pre-taught motion trajectory work file, collects the points that need to be photographed, saves the 2D pictures and 3D point cloud information of the corresponding points, and transmits them to the algorithm for train unqualified inspection.

[0051] Please refer to the steps shown in Figure 1 as shown in the steps:

[0052] S1: Set the preset parking point of the bottom inspection robot and the preset collection and photographing point of the mechanical arm of the bottom inspection robot under the preset parking point. Optionally, when setting the preset parking point of the bottom inspection robot, record the position of the chassis of the bottom inspection robot relative to the preset starting point of the bottom inspection robot for setting the preset parking point.

[0053] In a specific implementation, first, a 3D scanning device is used to model the bogie and tunnel at the bottom of the train (an example of the application is a motor car), and a 3D model of the train bottom and tunnel is generated. Then, teaching is performed by moving the bottom inspection robot to the train bottom position that needs to be inspected, and recording the position of the chassis of the bottom inspection robot relative to the preset starting point of the bottom inspection robot at this time as the basis for the preset parking point through navigation, odometer, etc. Further, the mechanical arm is moved to the specified collection point by manual dragging for recording. Finally, the preset parking point of the bottom inspection robot and the preset collection and photographing point of all mechanical arms under the preset parking point are obtained. When actual inspection is performed subsequently, the bottom inspection robot will directly park at the preset parking point and the mechanical arm will collect and photograph according to the preset collection and photographing point.

[0054] S2: When the bottom inspection robot reaches the preset parking point, the actual three-dimensional point cloud information of the train bottom bogie is obtained by a precision scanning camera, the actual three-dimensional point cloud information and the pre-collected teaching three-dimensional point cloud information are registered by a point cloud registration algorithm, and the deviation RT of the bottom inspection robot between the preset parking point and the teaching parking point is obtained.

[0055] In some embodiments of this application, the undercarriage inspection robot is equipped with a correction point, which is a camera position fixed relative to the robot arm chassis. At this point, the point cloud acquired by the precision scanning camera has a wide range and many features. When the undercarriage inspection robot reaches the parking point again, the robot arm moves to this point, and registering the acquired 3D point cloud with the 3D point cloud from the teaching phase directly reflects the deviation between the current parking point and the parking point from the teaching phase. Based on this, in a specific implementation, the robot arm is first moved to a correction point to acquire the actual 3D point cloud information of the current undercarriage bogie at the correction point using the precision scanning camera.

[0056] Specifically, the deviation RT is a 4x4 coordinate transformation matrix.

[0057] Specifically, a greedy algorithm is used to find the correspondence between two pairs of points in the point cloud: the current tunnel information and the tunnel model, and the current vehicle bottom information and the vehicle bottom model. The rotation relationship parameter R and the translation relationship parameter t are then calculated, and point cloud registration is performed based on the rotation relationship parameter R and the translation relationship parameter t.

[0058] In practice, when the bottom inspection robot performs automatic data collection, it first reaches the designated preset parking point using navigation algorithms and SLAM positioning. Due to errors caused by navigation algorithms and changes in the surrounding environment, it cannot be guaranteed that the preset parking point reached based on the robot's position relative to its preset starting point will perfectly match the parking point relative to the vehicle's underside during teaching. Therefore, after the bottom inspection robot reaches the parking point, it first collects the current information of the vehicle's underside and the tunnel using distance sensors (LiDAR and depth camera), and then matches it with the pre-collected tunnel model and vehicle underside model, i.e., the 3D models of the vehicle underside and tunnel mentioned in step S1, using an algorithm to finally obtain the position information of the robotic arm on the vehicle underside and in the tunnel.

[0059] In practice, during the teaching process, the robotic arm records the Cartesian coordinates (including position and orientation) of each preset image acquisition point in the coordinate system with the base of the robotic arm. When the bottom inspection robot reaches the preset parking point again, there is an error between the actual parking point reached relative to the preset starting point of the bottom inspection robot and the parking point relative to the bottom of the vehicle during the teaching process. If the robotic arm is not used for correction, the acquired image will have the same deviation as the standard image.

[0060] Furthermore, in specific implementation, taking the example that the bottom inspection robot deviated by 10cm from the parking point when it returned to the parking point during teaching, when the robotic arm starts the data collection task, the images collected at each photo point will also have the same deviation from the standard image.

[0061] In some embodiments of this application, the robotic arm is first moved to the correction point; specifically, the robotic arm makes small movements without colliding with the surrounding environment, and the actual three-dimensional point cloud information of the current undercarriage bogie is obtained by a fine scanning camera.

[0062] The point cloud registration algorithm is used to register the actual 3D point cloud information currently acquired with the 3D point cloud information recorded during teaching, so as to obtain the deviation RT between the current position of the bottom inspection robot under the bogie and the position during teaching. Specifically, the deviation RT is a 4x4 coordinate transformation matrix.

[0063] In some embodiments of this application, the point cloud registration problem is described as follows:

[0064]

[0065] In the formula, P s and P t These are the corresponding points in the source point cloud and the target point cloud.

[0066] In practice, a greedy algorithm is used to find the correspondence between points in two point clouds. Optionally, the closest point can be chosen as the corresponding point. Based on this, once the correspondence between points in the two point clouds is found, the parameters R and t are solved using LeastSquares.

[0067] Repeat the above steps alternately, iterating the calculation until convergence.

[0068] S3: Correct the preset acquisition and photography points according to the deviation RT to obtain the corrected acquisition and photography points. Perform path planning based on the corrected acquisition and photography points to generate the optimal planning path without collision.

[0069] In practice, all preset acquisition and photography points undergo RT coordinate transformation based on the robotic arm's base coordinate system according to the deviation RT.

[0070] In some embodiments of this application, such as Figure 2 As shown, the data required for coordinate transformation includes the model coordinate system, chassis coordinate system, robotic arm base coordinate system, and teaching photo point coordinates. The model coordinate system is a base coordinate system and will not change; the robotic arm base coordinate system is fixed relative to the chassis coordinate system; in specific implementation, in order to ensure that the image captured by the current photo point is highly coincident with the image captured by the photo point during teaching, the coordinates of the photo point relative to the model coordinate system are fixed.

[0071] In practice, when the parking position of the bottom inspection robot changes, that is, when the coordinates of the chassis change relative to the model coordinate system, the coordinates undergo translation and rotation, which is represented by the transformation matrix RT, or deviation RT. Since the base coordinate system of the robotic arm is fixed relative to the chassis coordinate system, in order to ensure that the current position of the photo-taking point coordinates relative to the model coordinate system and the teaching photo-taking point coordinates relative to the model coordinate system are consistent, the current photo-taking point position also undergoes the same RT transformation relative to the robotic arm base coordinate system.

[0072] In summary, the photographic points taken during the teaching demonstration all exhibit the same deviation RT at the location under the bogie. Therefore, the taught photographic points are subjected to the same RT coordinate transformation to obtain new, corrected photographic points.

[0073] Then, the robotic arm path planning algorithm is used to plan the photo-taking point after RT transformation.

[0074] Specifically, in some embodiments of this application, the robotic arm path planning algorithm includes the following when performing path planning:

[0075] Within the robotic arm's planning space, the location of the non-obstacle zone is determined, a random point is generated, and growth is performed.

[0076] Growth includes:

[0077] Using a random point as the target, traverse the existing nodes on the growth tree and calculate the distance from each existing node to the random point in order to filter out the closest point;

[0078] Using the line connecting the nearest point and a random point as the growth direction, grow from the nearest point towards the target point and generate growth points. Determine whether the growth point collides with an obstacle; if no collision occurs, add the growth point to the growth tree; if a collision occurs, remove the growth point.

[0079] After adding or removing growth points, new random points are generated, and growth is performed based on these new random points to obtain the optimal planned path.

[0080] In specific implementation, such as Figure 3 As shown in the figure, the starting point (left point) and ending point (right point) of the end axis of the robotic arm are shown in the planning space of the robotic arm, and obstacles are represented by black ellipses.

[0081] like Figure 4 As shown, in the planning space of the robotic arm, the location of the non-obstacle zone is determined, and a random point (a cross-shaped point) is randomly generated. Using the newly generated random point as the target, the existing nodes on the growth tree are traversed, the distance from each node to the random point is calculated, and the node with the smallest distance is selected as the nearest point. Specifically, at this point, only the starting point exists on the tree, so the starting point is directly selected as the nearest point.

[0082] like Figure 5 As shown, the growth direction is the line connecting the nearest point and a random point. Growth proceeds from the nearest point towards the target, with the growth length defined as the step size. Specifically, the step size for each growth is fixed, but can be set depending on the situation. A growth point of one step size is generated from the current nearest point (the starting point) along the growth direction. It is then checked whether the growth point collides with an obstacle. If not, the growth point is added to the tree; otherwise, it is removed.

[0083] like Figures 6-7 As shown, generate random points again and repeat the above steps to select the optimal planned path.

[0084] Please continue reading Figure 1 :

[0085] S4: Conduct a thorough inspection of the train based on the location of the correction data collection and the optimal planned route.

[0086] In practice, based on the correction and acquisition points and the optimal planned path, the corrected RGB image information and 3D point cloud information are obtained by taking pictures.

[0087] In summary, this application proposes a control method for a train undercarriage inspection robot, which abandons the previous method of recording the trajectory through manual teaching and repeatedly running the trajectory. Before manual teaching, each point requiring photography is recorded. When the undercarriage inspection robot reaches the designated position again, the robotic arm first acquires information about the surrounding tunnel and train undercarriage through environmental perception. This 3D point cloud information is added to the planning environment for collision detection. Since the position of the undercarriage inspection robot's chassis deviates from its position during teaching, it is matched with the standard point cloud data from teaching to calculate the RT (Real-Time) deviation between the robot's current position and its teaching position. Then, the taught points are also subjected to the same RT spatial transformation. The robotic arm, through collision detection and motion planning, plans a collision-free optimal motion path, which is sent to the robotic arm control system for control, and the robot reaches the designated point to collect data.

[0088] It should be understood that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order constraint on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the diagram may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0089] Please see Figure 8 One embodiment of this application provides a bottom inspection control device for a train bottom inspection robot, including a preset point setting module 10, a deviation RT calculation module 20, a correction path planning module 30, and a bottom inspection module 40; wherein,

[0090] The preset point setting module is used to set the preset parking point of the bottom inspection robot and the preset acquisition and photo taking point of the robot's robotic arm under the preset parking point;

[0091] The deviation RT calculation module is used to obtain the actual three-dimensional point cloud information of the current bogie under the vehicle through a fine scanning camera when the bottom inspection robot reaches the preset parking point. The actual three-dimensional point cloud information is registered with the pre-collected teaching three-dimensional point cloud information through a point cloud registration algorithm to obtain the deviation RT of the bottom inspection robot between the preset parking point and the teaching parking point.

[0092] The path planning module is used to correct the preset acquisition and photography points according to the deviation RT, to obtain the corrected acquisition and photography points, and to plan the path based on the corrected acquisition and photography points to generate the optimal planning path without collision.

[0093] The bottom inspection module performs a bottom inspection of the train based on the photo collection points for correction and the optimal planned path.

[0094] Specific limitations regarding the aforementioned train under-inspection robot control device can be found in the above description of the train under-inspection robot control method, and will not be repeated here. Each module in the aforementioned train under-inspection robot control device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the under-inspection robot in hardware form or independent of it, or they can be stored in the memory of the under-inspection robot in software form, so that the processor can call and execute the corresponding operations of each module.

[0095] In one embodiment, a bottom inspection robot is provided, which can be a non-railway-type train bottom inspection robot or a rail-type train bottom inspection robot. Specifically, the internal structure diagram of the bottom inspection robot can be as follows: Figure 9As shown. The bottom inspection robot includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores data. The network interface communicates with external terminals via a network connection. When the processor executes the computer program, it implements the above-described bottom inspection control method for a train bottom inspection robot. It includes: a memory and a processor; the memory stores the computer program; and the processor executes the computer program to implement any step in the above-described bottom inspection control method for a train bottom inspection robot.

[0096] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, can perform any step of the above-described train bottom inspection robot bottom inspection control method.

[0097] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as C, VHDL, Verilog, the object-oriented programming language Java, and the interpreted scripting language JavaScript.

[0098] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0099] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0100] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0101] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0102] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A control method for a train under-inspection robot, characterized in that, include: Set the preset parking point of the bottom inspection robot and the preset image capture point of the robot's robotic arm at the preset parking point; When the bottom inspection robot reaches the preset parking point, it acquires the actual three-dimensional point cloud information of the current undercarriage bogie through a fine scanning camera. The actual three-dimensional point cloud information is then registered with the pre-collected teaching three-dimensional point cloud information using a point cloud registration algorithm to obtain the deviation RT between the preset parking point and the teaching parking point of the bottom inspection robot. The preset acquisition and photography points are corrected according to the deviation RT to obtain the corrected acquisition and photography points. Path planning is performed based on the corrected acquisition and photography points to generate the optimal planning path without collision. Based on the aforementioned correction and image capture locations and the optimal planned path, a thorough inspection of the train is conducted. The step of correcting the preset acquisition and photography points according to the deviation RT to obtain the corrected acquisition and photography points further includes: performing RT coordinate transformation based on the robot arm base coordinate system on all the preset acquisition and photography points according to the deviation RT. The step of performing path planning based on the aforementioned correction and image acquisition points to generate a collision-free optimal planned path further includes: Within the planned space of the robotic arm, the location of the non-obstacle zone is determined, a random point is generated, and growth is performed; the growth includes: Using the random point as the target, traverse the existing nodes on the growth tree and calculate the distance from each existing node to the random point in order to filter out the closest point; Using the line connecting the nearest point and the random point as the growth direction, grow from the nearest point to the target point and generate growth points, and determine whether the growth points collide with obstacles. If no collision occurs, the growth point is added to the growth tree; If a collision occurs, the growth point is removed, a new random point is generated, and the growth is performed based on the new random point to obtain the optimal planned path.

2. The train under-inspection robot control method according to claim 1, characterized in that, The step of registering the actual 3D point cloud information and the pre-collected teaching 3D point cloud information using a point cloud registration algorithm to obtain the deviation RT between the preset parking point and the teaching parking point of the bottom inspection robot further includes: The deviation RT is a 4x4 coordinate transformation matrix.

3. The train under-inspection robot control method according to claim 1 or 2, characterized in that, The growth further includes: the growth step size is a fixed value.

4. The train under-inspection robot control method according to claim 1 or 2, characterized in that, The step of registering the actual 3D point cloud information and the pre-acquired teaching 3D point cloud information using a point cloud registration algorithm further includes: The correspondence between points on the actual 3D point cloud information and the teaching 3D point cloud information is found by a greedy algorithm. The rotation relationship parameter R and the translation relationship parameter t are further calculated. Point cloud registration is performed based on the rotation relationship parameter R and the translation relationship parameter t.

5. The train under-inspection robot control method according to claim 1, characterized in that, Set the preset parking point for the bottom inspection robot and the preset image capture point for the robot's robotic arm at the preset parking point: When setting the preset parking point of the bottom inspection robot, the position of the chassis of the bottom inspection robot relative to the preset starting point of the bottom inspection robot is recorded to set the preset parking point.

6. A control device for a train under-inspection robot, characterized in that, It includes a preset point setting module, a deviation RT calculation module, a correction path planning module, and a bottom inspection module; among which, The preset point setting module is used to set the preset parking point of the bottom inspection robot and the preset acquisition and photography point of the robotic arm of the bottom inspection robot under the preset parking point; The deviation RT calculation module is used to obtain the actual three-dimensional point cloud information of the current undercarriage bogie through a fine scanning camera when the undercarriage inspection robot reaches the preset parking point. The actual three-dimensional point cloud information is then registered with the pre-collected teaching three-dimensional point cloud information through a point cloud registration algorithm to obtain the deviation RT of the undercarriage inspection robot between the preset parking point and the teaching parking point. The deviation correction path planning module is used to correct the preset acquisition and photography points according to the deviation RT, to obtain the corrected acquisition and photography points, and to perform path planning based on the corrected acquisition and photography points to generate a collision-free optimal planning path. The bottom inspection module performs a bottom inspection of the train based on the correction acquisition and photography points and the optimal planned path. The deviation correction path planning module corrects the preset acquisition and photography points according to the deviation RT, and the corrected acquisition and photography points further include: performing RT coordinate transformation based on the robot arm base coordinate system on all the preset acquisition and photography points according to the deviation RT. The path planning module for correction further includes the following: Based on the correction data acquisition points, the module performs path planning to generate a collision-free optimal planned path. Within the planned space of the robotic arm, the location of the non-obstacle zone is determined, a random point is generated, and growth is performed; the growth includes: Using the random point as the target, traverse the existing nodes on the growth tree and calculate the distance from each existing node to the random point in order to filter out the closest point; Using the line connecting the nearest point and the random point as the growth direction, grow from the nearest point to the target point and generate growth points, and determine whether the growth points collide with obstacles. If no collision occurs, the growth point is added to the growth tree; If a collision occurs, the growth point is removed, a new random point is generated, and the growth is performed based on the new random point to obtain the optimal planned path.

7. A bottom inspection robot, characterized in that, include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and configured to be executed by the processor to implement the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, It stores a computer program thereon; the computer program is executed by a processor to implement the method as described in any one of claims 1-5.

Images