Inspection Robot Charging Control Method and System Based on Point Cloud Matching

Abstract

The present invention discloses an inspection robot charging control method and system based on point cloud matching. The method includes the steps: after receiving a charging instruction, the inspection robot runs to a charging point; collecting real-time laser point cloud data of the inspection robot in a stationary state; performing point cloud matching on the collected real-time laser point cloud data and the pre-stored laser data of the charging point to obtain pose difference information of the inspection robot; performing pose adjustment according to the pose difference information; after the pose of the inspection robot is adjusted, retreating to a target position set between the charging pile and the target charging point, inserting the charging pin of the inspection robot into the charging pile, powering on the charging pile, and starting to charge. The present invention does not rely on the internal features of the charging room, performs matching positioning based on the point cloud data of the charging point, is simple to operate, has accurate positioning, and improves the charging success rate.

Description

Technical Field

[0001] The present invention belongs to the technical field of charging for inspection robots, and particularly relates to a charging control method and system for inspection robots based on point cloud matching. Background Art

[0002] Currently, inspection robots have been increasingly applied in fields such as security inspection, power inspection, indoor inspection, and computer room inspection. Inspection robots can usually navigate and position autonomously and return to the charging room to charge automatically when the battery runs low. The key factor for the successful charging of inspection robots is to accurately dock with the charging pile, so it is required that the pose of the robot be as precise as possible. Most current inspection robot chargings rely on internal features in the charging room for matching and positioning, which is prone to losing the positioning, and the configuration is cumbersome with frequent problems. Summary of the Invention

[0003] Technical Objective: Aiming at the above technical problems, the present invention provides a charging method and system for inspection robots based on point cloud matching that does not rely on the charging room. It does not rely on internal features of the charging room, performs matching and positioning based on the point cloud data of the charging point, is simple to operate, has precise positioning, and can improve the charging success rate.

[0004] Technical Solution: To achieve the above technical objective, the present invention adopts the following technical solution:

[0005] A charging control method for inspection robots based on point cloud matching, with the execution subject being the inspection robot, includes the steps of:

[0006] Receiving a charging instruction, obtaining the target charging point information in the charging instruction, and moving to the target charging point;

[0007] Receiving pose difference information and adjusting the pose according to the pose difference information;

[0008] After the pose is adjusted, retreat to the target position set between the target charging point and the charging pile, extend the charging pin and insert it into the charging pile to start charging;

[0009] Receiving an end charging instruction, retracting the charging pin, returning to the target charging point, and ending the charging;

[0010] Among them, the pose difference information is obtained by collecting the real-time laser point cloud data of the inspection robot in a stationary state at the charging point and performing point cloud matching between the real-time laser point cloud data and the pre-stored laser data of the charging point. The real-time laser point cloud data is the surrounding environment information scanned by the inspection robot at the current charging point pose, and the pre-stored laser data of the charging point is the surrounding environment information scanned by the laser of the inspection robot located at the charging point when collecting the map points.

[0011] A charging control method for an inspection robot based on point cloud matching, with the execution subject being the inspection robot server, including the steps:

[0012] Send a charging instruction to the inspection robot;

[0013] When the inspection robot runs to the set target charging point, collect the real-time laser point cloud data of the inspection robot in a stationary state;

[0014] Perform point cloud matching on the collected real-time laser point cloud data and the pre-stored laser data of the charging point. The pre-stored laser data of the charging point is the surrounding environment information scanned by the laser of the inspection robot at the charging point when collecting the map points, and obtain the pose difference information between the current pose of the inspection robot and the pose of the inspection robot when collecting the map points;

[0015] After the pose of the inspection robot is adjusted, retreat to the target position set between the target charging point and the charging pile. After the charging pin of the inspection robot is inserted into the charging pile, power on the charging pile.

[0016] Preferably, after the charging pile is powered on, perform the following steps:

[0017] Detect the current or voltage of the charging pin of the inspection robot, and judge whether the data of the current or voltage is normal. If the data is normal, judge that the charging is successful; otherwise, judge that the charging is unsuccessful;

[0018] Count and judge the number of times the inspection robot fails to charge successfully on the target charging pile. If it is less than or equal to the preset number of times, send a charging instruction to the inspection robot; otherwise, judge that the charging fails, issue a warning, and end the charging.

[0019] Preferably, the inspection robot server performs the following steps to end the charging:

[0020] Send an end charging instruction to the inspection robot;

[0021] Judge whether the inspection robot is in the target position. If so, cut off the power of the charging pile. After the inspection robot retracts the charging pin and returns to the charging point, end the charging; otherwise, end the charging.

[0022] Preferably, collect multiple frames of real-time laser point cloud data, and perform point cloud matching frame by frame with the pre-stored laser data of the charging point using the PLICP algorithm, including the steps:

[0023] Project the current frame data into the reference frame coordinate system according to the initial pose;

[0024] For the points in the current frame, find the two nearest points in the reference frame using the nearest neighbor rule;

[0025] Connect the two closest points, calculate the distance from the points in the current frame to the connection line as the error, and remove the points with excessive error;

[0026] Solve the rotation matrix and translation matrix according to the minimized error function;

[0027] Use the translation and rotation parameters obtained in the previous step for the current frame to obtain a new set of transformed points;

[0028] If the average distance between the new set of transformed points and the reference point set satisfies that it is less than a given threshold, stop the iterative calculation; otherwise, use the new set of transformed points as the new current frame and continue the iteration until the requirements of the objective function are met.

[0029] Preferably, the calculation formula of the objective function is as follows:

[0030]

[0031] Among them, P i represents the i-th sampling point, is a matching point that is the nearest neighbor of the sampling point in the reference frame coordinate system, n i is the normal vector of the two nearest neighbor matching points, R k+1, t k+1 represent the rotation vector and the translation vector respectively.

[0032] An inspection robot charging system based on point cloud matching includes an inspection robot, and the inspection robot is provided with a control instruction receiving module and a pose adjustment driving module.

[0033] The control instruction receiving module is used to receive the charging instruction and the end charging instruction sent externally;

[0034] The pose adjustment driving module is used to receive the pose difference information and perform pose adjustment on the inspection robot according to the pose difference information;

[0035] The navigation module is used to move to the target charging point according to the target charging point information in the charging instruction, or drive the inspection robot to retreat to the target position set between the charging pile and the target charging point after the pose is adjusted;

[0036] The charging needle control module is used to control the charging needle to extend and insert into the charging pile, or disengage from the charging pile.

[0037] Preferably, it includes an inspection robot server, and the inspection robot server is provided with a charging point laser data packet recording module, a laser static matching module, a real-time data acquisition module and a charging control module. Among them,

[0038] The charging control module is used to send the charging instruction and the end charging instruction to the inspection robot;

[0039] A charging point laser data packet recording module, which is used to pre-store the charging point laser data when the inspection robot is located at the charging point and in a stationary state when a new charging pile and charging point are created. The pre-stored charging point laser data is the surrounding environment information scanned by the laser of the inspection robot at the charging point when collecting map points;

[0040] A real-time data acquisition module, which is used to collect the real-time laser point cloud data of the inspection robot in a stationary state when the inspection robot runs to a set target charging point;

[0041] A laser static matching module, which is used to perform point cloud matching between the collected real-time laser point cloud data and the pre-stored charging point laser data to obtain the pose difference information between the current pose of the inspection robot and the pose of the inspection robot when collecting map points. The real-time laser point cloud data is the surrounding environment information scanned by the inspection robot at the current charging point pose;

[0042] A sending module, which is used to send the pose difference information obtained by point cloud matching to the inspection robot. The pose difference information is obtained by collecting the real-time laser point cloud data of the inspection robot in a stationary state at the charging point and performing point cloud matching between the real-time laser point cloud data and the pre-stored charging point laser data.

[0043] Advantageous effects: Due to the adoption of the above technical solutions, the present invention has the following advantageous effects:

[0044] The present invention provides an inspection robot charging method based on point cloud matching. This method does not rely on a charging room and is relatively simple, only requiring the installation of a charging pile. When establishing a charging point, the environmental information scanned by the laser is stored. When the robot charges, it first reaches the charging point relying on the pose information, and then the current frame point cloud information and the point cloud information stored in the data packet are statically matched through point-line ICP to obtain the deviation information of the robot pose. Combined with the position control of the embedded software, secondary fine-tuning is performed to enable the robot to obtain a more accurate pose, thereby improving the charging success rate. Description of the Drawings

[0045] Figure 1 It is a flowchart of the method when the execution subject is the inspection robot for the inspection robot charging control method based on point cloud matching in Embodiment 1;

[0046] Figure 2 It is a flowchart of the method when the execution subject is the industrial control computer of the inspection robot for the inspection robot charging control method based on point cloud matching in Embodiment 1;

[0047] Figure 3 It is a flowchart of the working process of the charging point laser data recording module in Embodiment 2;

[0048] Figure 4 It is the flowchart of the working process of the laser static matching module in Embodiment 2;

[0049] Figure 5 It is the flowchart of the working process of the charging control module for charging in Embodiment 2;

[0050] Figure 6 It is the flowchart of the working process of the charging control module to end charging in Embodiment 2;

[0051] Figure 7 It is the flowchart of the point-to-point distance calculation method in the point cloud matching algorithm;

[0052] Figure 8 It is the flowchart of the point-to-line distance calculation method in the point cloud matching algorithm. Specific implementation manners

[0053] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0054] Embodiment 1

[0055] This embodiment proposes a charging control method for an inspection robot based on point cloud matching. This method does not rely on a charging room and is relatively simple, only requiring the installation of a charging pile. When establishing a charging point, the environmental information scanned by the laser is stored. When the robot charges, it first reaches the charging point relying on the pose information, and then the current frame point cloud information and the point cloud information stored in the data packet are statically matched through point-line ICP to obtain the deviation information of the robot pose. Combined with the position control of the embedded software, secondary fine-tuning is performed to enable the robot to obtain a more accurate pose, thereby improving the charging success rate.

[0056] The charging point referred to in the present invention is a transition position established around the charging pile. For example, the charging point is set in front of the charging pile. There is a target position between the charging pile and the charging point. The inspection robot first runs to the charging point, adjusts its pose at the charging point and then retreats to the target position, extends the charging needle at the target position and inserts it into the charging pile, and then starts charging.

[0057] As Figure 1 shown, when the method execution subject is an inspection robot, it includes the steps:

[0058] Receive a charging instruction, obtain the target charging point information in the charging instruction, and move to the set target charging point;

[0059] Receive the pose difference information and perform pose adjustment according to the pose difference information;

[0060] After the pose is adjusted, retreat to the target position set between the target charging point and the charging pile, extend the charging needle and insert it into the charging pile, and start charging;

[0061] Receive the end - charging instruction, retract the charging pin, return to the charging point in front of the target charging pile, and end the charging;

[0062] Among them, the pose difference information is obtained by collecting the real - time laser point - cloud data of the inspection robot in a stationary state at the charging point and performing point - cloud matching between the real - time laser point - cloud data and the pre - stored laser data of the charging point. The real - time laser point - cloud data is the surrounding environment information scanned by the inspection robot in the current charging - point pose, and the pre - stored laser data of the charging point is the surrounding environment information scanned by the laser of the inspection robot located at the charging point when collecting the map points.

[0063] As Figure 2 shown, the execution subject of the method is the robot server, including the steps:

[0064] Send a charging instruction to the inspection robot;

[0065] When the inspection robot runs to the set target charging point, collect the real - time laser point - cloud data of the inspection robot in a stationary state;

[0066] Perform point - cloud matching between the collected real - time laser point - cloud data and the pre - stored laser data of the charging point. The pre - stored laser data of the charging point is the surrounding environment information scanned by the laser of the inspection robot located at the charging point when collecting the map points, and obtain the pose difference information between the current pose of the inspection robot and the pose of the inspection robot when collecting the map points;

[0067] After the pose of the inspection robot is adjusted, retreat to the target position set between the target charging pile and the charging point. After the charging pin of the inspection robot is inserted into the charging pile, power on the charging pile.

[0068] When the charging pile is powered on, the following steps can be executed:

[0069] Detect the current or voltage of the charging pin of the inspection robot, and judge whether the data of the current or voltage is normal. If the data is normal, judge that the charging is successful; otherwise, judge that the charging is unsuccessful;

[0070] Count and judge the number of times that the inspection robot fails to charge successfully on the target charging pile. If it is less than or equal to the preset number of times, send a charging instruction to the inspection robot; otherwise, judge that the charging fails, issue a warning, and end the charging.

[0071] The robot server executes the following end - charging steps:

[0072] Send an end - charging instruction to the inspection robot;

[0073] Judge whether the inspection robot is in the target position. If so, cut off the power of the charging pile. After the inspection robot retracts the charging pin and returns to the charging point, end the charging; otherwise, end the charging.

[0074] Embodiment 2

[0075] This embodiment provides an inspection robot charging system based on point cloud matching, including an inspection robot, which is provided with a control instruction receiving module and a pose adjustment driving module.

[0076] The control instruction receiving module is used to receive the charging instruction and the end charging instruction sent externally.

[0077] The pose adjustment driving module is used to receive the pose difference information and adjust the pose of the inspection robot according to the pose difference information.

[0078] The navigation module is used to move to the charging point set in front of the target charging pile according to the target charging pile information in the charging instruction, or drive the inspection robot to retreat to the target position set between the target charging pile and the charging point after the pose is adjusted.

[0079] The charging pin control module is used to control the charging pin to extend and insert into the charging pile, or disengage from the charging pile.

[0080] The inspection robot charging system based on point cloud matching further includes a robot server, which is provided with a charging point laser data packet recording module, a laser static matching module, a real-time data acquisition module and a charging control module. Among them,

[0081] The charging point laser data packet recording module is used to pre-store the charging point laser data when the inspection robot is located at the charging point and in a stationary state when a new charging pile and charging point are created. The pre-stored charging point laser data is the surrounding environment information scanned by the laser of the inspection robot when collecting map points at the charging point.

[0082] The real-time data acquisition module is used to collect the real-time laser point cloud data of the inspection robot in a stationary state when the inspection robot runs to the set target charging point.

[0083] The laser static matching module is used to perform point cloud matching on the collected real-time laser point cloud data and the pre-stored charging point laser data to obtain the pose difference information between the current pose of the inspection robot and the pose of the inspection robot when collecting map points. The real-time laser point cloud data is the surrounding environment information scanned by the inspection robot at the current charging point pose.

[0084] The sending module is used to send the pose difference information obtained by point cloud matching to the inspection robot. The pose difference information is obtained by collecting the real-time laser point cloud data of the inspection robot in a stationary state at the charging point and performing point cloud matching on the real-time laser point cloud data and the pre-stored charging point laser data.

[0085] The following is combined withFigures 3 to 6 , the various modules in the robot industrial control machine are described in detail.

[0086] 1. Record the laser data packet of the charging point

[0087] When establishing a charging point, the laser scan data at the charging point position is stored, that is, the laser data packet of the charging point is recorded. The laser data packet of the charging point includes the point cloud information of the charging point position for static matching operations during later charging.

[0088] Collect map points through the host-side site-building tool. When the charging point position information is collected, the robot is at the charging point. At this time, click the button to record the static aurora data packet on the page of the site-building tool. After receiving the instruction, the laser static matching module will read 10 frames of surrounding environment point cloud data scanned by the laser at the current position and store it in the robot industrial control machine. Subsequently, the laser static matching module provides the original data. The process of recording the laser data packet of the charging point is as Figure 3 shown. The inspection robot executes the autonomous charging task and navigates to the charging point, starts the laser static matching module, matches the current frame of laser point cloud data with the previously stored original data packet, and obtains the difference between the pose of the robot when collecting the charging point position and the current pose of the robot. During the matching process, the pre-stored point cloud information of the charging point is used as the true value, and the real-time laser point cloud data is based on the true value, continuously adjusting the pose of the inspection robot so that the real-time laser point cloud data coincides with the pre-stored point cloud data. Click the button to record the static laser data packet on the site-building tool side. After receiving the instruction, the laser static matching module will read 10 frames of laser data and store it in the robot industrial control machine.

[0089] 2. Laser static matching module

[0090] Obtain the point cloud information of the pre-recorded charging point position, perform point-line ICP matching with the current frame of point cloud data, and obtain the difference between the pose of the robot when collecting the charging point position and the current pose of the robot, as the basis for the robot to adjust its pose.

[0091] The static matching of the laser refers to the matching of two or more frames of point clouds of the laser when the robot is in a stationary state. The matching of point clouds is to find the relationship between the corresponding points of the two sets of point clouds, so as to solve the rotation and translation relationship of the two sets of point clouds.

[0092] In the present invention, the laser static matching used is the optimized method of ICP iterative closest point, the PLICP (point-line ICP) method, to match two frames of laser data. ICP point cloud matching is an optimal registration method based on the least squares. The input of the algorithm is the reference point cloud, the target point cloud and the stop iteration criterion, and the output of the algorithm is the transformation matrix, that is, the rotation and translation amount.

[0093] PLICP is an improved ICP algorithm. The ICP algorithm finds the nearest neighbor point to solve the point-to-point distance, while the PLICP algorithm solves the point-to-line distance. Combining Figure 7 and Figure 8 as shown, the algorithm flow is as follows:

[0094] (1) Project the current frame data into the reference frame coordinate system according to the initial pose;

[0095] (2) For the points in the current frame, find the two nearest points in the reference frame using the nearest neighbor rule;

[0096] (3) Connect the two nearest points, calculate the distance from the point in the current frame to the line as the error, and remove the points with too large errors;

[0097] (4) Solve the rotation matrix and translation matrix according to the minimization error function;

[0098] (5) Use the translation and rotation parameters obtained in the previous step for the current frame to obtain a new set of transformed points;

[0099] (6) If the average distance between the new set of transformed points and the reference point set satisfies that it is less than a given threshold, stop the iterative calculation; otherwise, use the new set of transformed points as the new current frame and continue the iteration until the requirements of the objective function are met.

[0100] The calculation formula of the objective function is as follows:

[0101]

[0102] where, P i represents the i-th sampling point, is a matching point that is the nearest neighbor of the sampling point in the reference frame coordinate system, n i is the normal vector of the two nearest neighbor matching points, R k+1, t k+1 represent the rotation vector and translation vector respectively.

[0103] 3. Charging control module

[0104] After receiving the return charging instruction, the inspection robot autonomously navigates to the charging point relying on the pose information. The pose information refers to the pose information of the robot, including the translation amount and rotation amount, and starts the charging process.

[0105] The first step of the charging process is to call the static matching interface through the service mode of ROS to obtain the differences in the x-direction, y-direction, and yaw angle. Since the angle error directly affects the deviation in the front, back, left, and right directions, the yaw angle, i.e., the orientation, is adjusted first. After the orientation deviation meets the error redundancy, the y-direction and x-direction are adjusted in sequence until all pose errors meet the requirements, and it is considered that the robot pose adjustment is completed.

[0106] Then the robot starts to rely on the odometer to back up to the front of the charging pile. When the proximity switch touches the charging pile, the robot extends the charging arm and starts to power on the charging pile. After the power-on is successful, it means the charging is successful. After the power-on fails, the robot will return to the charging point again to execute the charging operation. Repeating the above behavior, if the number of failure times exceeds the limit, the charging failure will be reported.

[0107] After the inspection robot receives the instruction to end charging, it first judges whether the robot itself is in the charging area. If it is confirmed that it is in the charging area, the power-off instruction for the charging pile will be issued, then the charging arm will be retracted, and it will navigate to the charging point to complete the end-charging operation. If the end-charging instruction is received outside the charging area, the robot will directly end the charging operation.

[0108] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A charging control method for an inspection robot based on point cloud matching, characterized in that, The execution entity is the inspection robot, including the steps: Receive a charging instruction, obtain the target charging point information in the charging instruction, and move to the target charging point; Receive the pose difference information and perform pose adjustment according to the pose difference information; After the pose is adjusted, retreat to the target position set between the target charging point and the charging pile, extend the charging needle and insert it into the charging pile to start charging; Receive an end charging instruction, retract the charging needle, return to the target charging point, and end charging; Among them, the pose difference information is obtained by collecting the real-time laser point cloud data of the inspection robot in a stationary state at the charging point and performing point cloud matching between the real-time laser point cloud data and the pre-stored laser data of the charging point. The real-time laser point cloud data is the surrounding environment information scanned by the inspection robot at the current charging point pose, and the pre-stored laser data of the charging point is the surrounding environment information scanned by the laser of the inspection robot located at the charging point when collecting the map points; Collect multiple frames of real-time laser point cloud data and perform point cloud matching with the pre-stored laser data of the charging point frame by frame using the PLICP algorithm, including the steps: Project the current frame data into the reference frame coordinate system according to the initial pose; For the points in the current frame, find the two closest points in the reference frame using the nearest neighbor rule; Calculate the distance from the points in the current frame to the line connecting the two closest points as the error, and remove the points with too large errors; Solve the rotation matrix and translation matrix according to the minimization error function; Use the translation and rotation parameters obtained in the previous step for the current frame to obtain a new set of transformed points; If the new set of transformed points and the reference point set satisfy that the average distance between the two point sets is less than a given threshold, stop the iterative calculation, otherwise the new set of transformed points is used as the new current frame to continue the iteration until the requirements of the objective function are met; The calculation formula of the objective function is as follows: Among them, P i represents the i-th sampling point, is a matching point that is the nearest neighbor of the sampling point in the reference frame coordinate system, and n i is the normal vector of two nearest neighbor matching points, and R k+1 , t k+1 respectively represent the rotation vector and the translation vector.

2. A charging control method for an inspection robot based on point cloud matching, characterized in that, The execution entity is the inspection robot server, including the steps: Send a charging instruction to the inspection robot; When the inspection robot runs to the set target charging point, collect the real-time laser point cloud data of the inspection robot in a stationary state; Perform point cloud matching between the collected real-time laser point cloud data and the pre-stored laser data of the charging point. The pre-stored laser data of the charging point is the surrounding environment information scanned by the laser of the inspection robot located at the charging point when collecting the map points, and obtain the pose difference information between the current pose of the inspection robot and the pose of the inspection robot when collecting the map points; After the pose of the inspection robot is adjusted, retreat to the target position set between the target charging point and the charging pile. After the charging needle of the inspection robot is inserted into the charging pile, power on the charging pile; Collect multiple frames of real-time laser point cloud data and perform point cloud matching with the pre-stored laser data of the charging point frame by frame using the PLICP algorithm, including the steps: Project the current frame data into the reference frame coordinate system according to the initial pose; For the points in the current frame, find the two closest points in the reference frame using the nearest neighbor rule; Calculate the distance from the points in the current frame to the line connecting the two closest points as the error, and remove the points with too large errors; Solve the rotation matrix and translation matrix according to the minimization error function; Use the translation and rotation parameters obtained in the previous step for the current frame to obtain a new set of transformed points; If the average distance between the new set of transformed points and the reference point set is less than a given threshold, stop the iterative calculation; otherwise, use the new set of transformed points as the new current frame and continue the iteration until the requirements of the objective function are met. The calculation formula of the objective function is as follows: Among them, P i represents the i-th sampling point, is a matching point that is the nearest neighbor of the sampling point in the reference frame coordinate system, and n i is the normal vector of the two nearest neighbor matching points, and R k+1 , t k+1 represent the rotation vector and the translation vector respectively.

3. A patrol robot charging control method based on point cloud matching according to claim 2, characterized in that, After the charging pile is powered on, perform the following steps: Detect the current or voltage of the charging pin of the inspection robot, and determine whether the data of the current or voltage is normal. If the data is normal, determine that the charging is successful; otherwise, determine that the charging is unsuccessful. Count the number of times the inspection robot fails to charge successfully at the target charging pile. If it is less than or equal to the preset number of times, send a charging instruction to the inspection robot; otherwise, determine that the charging fails, issue a warning, and end the charging.

4. The charging control method of the inspection robot based on point cloud matching according to claim 2, characterized in that The inspection robot server performs the following steps to end the charging: Send an end charging instruction to the inspection robot; Determine whether the inspection robot is at the target position. If so, cut off the power of the charging pile. After the inspection robot retracts the charging pin and returns to the charging point, end the charging; otherwise, end the charging.

5. A patrol robot charging system based on point cloud matching, characterized in that: Including an inspection robot, the inspection robot is provided with a control instruction receiving module and a pose adjustment driving module. The control instruction receiving module is used to receive the charging instruction and the end charging instruction sent externally. The pose adjustment driving module is used to receive the pose difference information and adjust the pose of the inspection robot according to the pose difference information. The navigation module is used to move to the target charging point according to the target charging point information in the charging instruction, or drive the inspection robot to retreat to the target position set between the charging pile and the target charging point after the pose is adjusted. The charging pin control module is used to control the charging pin to extend and insert into the charging pile, or disengage from the charging pile. Collect multiple frames of real-time laser point cloud data, and perform point cloud matching frame by frame using the PLICP algorithm with the pre-stored laser data of the charging point, including the steps of: Project the current frame data into the reference frame coordinate system according to the initial pose. For the points in the current frame, find the two closest points in the reference frame using the nearest neighbor rule. Calculate the distance from the points in the current frame to the line connecting the two closest points as the error, and remove the points with too large errors. Solve the rotation matrix and translation matrix according to the minimum error function. Use the translation and rotation parameters obtained in the previous step for the current frame to obtain a new set of transformed points. If the average distance between the new set of transformed points and the reference point set is less than a given threshold, stop the iterative calculation; otherwise, use the new set of transformed points as the new current frame and continue the iteration until the requirements of the objective function are met. The calculation formula of the objective function is as follows: Among them, P i represents the i-th sampling point, is a matching point that is the nearest neighbor of the sampling point in the reference frame coordinate system, and n i is the normal vector of two nearest neighbor matching points, and R k+1 , t k+1 represent the rotation vector and the translation vector respectively.

6. The inspection robot charging system based on point cloud matching according to claim 5, characterized in that: Including an inspection robot server, the inspection robot server is provided with a charging point laser data packet recording module, a laser static matching module, a real-time data acquisition module, and a charging control module. Among them, The charging control module is used to send a charging instruction and an end charging instruction to the inspection robot. The charging point laser data packet recording module is used to pre-store the laser data of the charging point when the inspection robot is at the charging point and in a stationary state when a new charging pile and charging point are created. The pre-stored laser data of the charging point is the surrounding environment information scanned by the laser of the inspection robot when collecting the map points at the charging point. A real-time data acquisition module, which is used to collect real-time laser point cloud data of the inspection robot in a stationary state when the inspection robot runs to a set target charging point; A laser static matching module, which is used to perform point cloud matching on the collected real-time laser point cloud data and the pre-stored laser data of the charging point to obtain pose difference information between the current pose of the inspection robot and the pose of the inspection robot when collecting map points. The real-time laser point cloud data is the surrounding environment information scanned by the inspection robot at the current charging point pose; A sending module, which is used to send the pose difference information obtained by point cloud matching to the inspection robot. The pose difference information is obtained by collecting real-time laser point cloud data of the inspection robot in a stationary state at the charging point and performing point cloud matching on the real-time laser point cloud data and the pre-stored laser data of the charging point.

Images