Integrated positioning and navigation method for urban rail vehicle inspection robots

Abstract

This invention proposes a comprehensive positioning and navigation method for urban rail vehicle inspection robots, belonging to the field of intelligent positioning and navigation. It can solve the technical problems of low compensation accuracy, susceptibility to external interference, and low maintenance efficiency of existing positioning and navigation methods. The method includes: (1) the laser radar detects the reflective sticker, compares its actual position on the map with the current position of the inspection robot, and compensates the X and Y axis coordinates of the inspection robot; the positioning camera detects specific parts under the train and feeds back their X and Y axis coordinates to compensate the X and Y axis coordinates of the inspection robot again; (2) the Z axis coordinate of the inspection robot is height-compensated according to the Z value of the specific parts under the train fed back by the positioning camera; the QR code detection camera calculates the height difference, and then compares the height difference with the height difference calculated during teaching at the current inspection point to compensate the height of the inspection robot's Z axis coordinate. This invention has the characteristics of high positioning accuracy, strong anti-interference, and wide applicability.

Description

Technical Field

[0001] This invention belongs to the field of intelligent positioning and navigation technology, and in particular relates to a comprehensive positioning and navigation method for an urban rail vehicle inspection robot. Background Technology

[0002] To automate and intelligentize inspection operations in the rail transit industry, the first step is to solve the train positioning problem. Currently, trains are manually parked on the tracks to be inspected, but due to various factors, the stopping position can vary each time, with a stopping error range of no more than 1 meter. This necessitates that intelligent inspection robots accurately locate the train's position before performing inspection tasks.

[0003] The movement of intelligent inspection robots can be divided into two scenarios. One involves deployment and teaching the robot, where path planning is performed on the robotic arm based on the distribution of vehicle components. The robot records relevant parameters at different inspection points (mainly including parameters of each axis of the robotic arm, the height of the lifting column, the height of the QR code, etc.; each stopping position is called an inspection point). The other scenario involves the robot executing tasks according to the planned path. However, this execution is affected by factors such as the train's stopping position and wheel wear, resulting in differences from the deployment and teaching state. Nevertheless, using intelligent inspection robots for train positioning still faces numerous challenges.

[0004] For example, because most of the columns on both sides of the trench track are of the same size, and the last 10 meters or so of the trench track are entirely walls with almost no obvious features (i.e., the environment is highly similar), the inspection robot cannot achieve high-precision trackless navigation and positioning. Let's define the robot's forward direction as the X-axis and the direction inside the trench track as the Y-axis (i.e., perpendicular to the robot's forward direction). If the inspection robot cannot accurately position itself in the X and Y directions, the posture of the vehicle parts photographed during inspection will differ significantly from the posture photographed during deployment and teaching, and may even result in parts being outside the camera's field of view. This prevents the visual algorithm from effectively processing the image. Currently, the most common technical solution for X-axis and Y-axis errors caused by inaccurate positioning is to increase the camera's field of view to ensure that vehicle parts appear as close to the center of the camera's view as possible, while also improving the compatibility of the visual algorithm. However, this solution cannot guarantee that all vehicle parts to be inspected will be within the camera's field of view. Furthermore, visual image processing algorithms struggle to effectively process the various postures of vehicle parts and cannot obtain the true state of the vehicle parts from the photographic data. At the same time, the complexity of the algorithm also increases the image processing time, which seriously affects the efficiency of maintenance.

[0005] To facilitate drainage, the subway maintenance depot's trench tracks are designed to be uneven, with undulations. This results in the trench track surfaces not being on a single horizontal plane; the maximum height difference measured across the entire trench track can reach 10cm. After subway vehicles have been running on the mainline for a period of time, the wheels will wear down to some extent, causing the overall height of the train to decrease. The maximum height difference between new and old trains can reach 2-3cm. In addition, maintenance work on the trains also causes changes in height. If inspection operations are carried out at the initial deployment and teaching height of the robotic arm, the accuracy in the height direction cannot meet the requirements. The decrease in train height greatly increases the risk of the robotic arm hitting undercarriage components during inspection. Therefore, it is required that the inspection robot perform real-time height (in the Z-axis direction) compensation during inspection. Currently, the most common technical solution for Z-axis height compensation is to repave the trench track surface to ensure flatness. This necessitates excavation work on the trench tracks, which, while ensuring ground flatness, hinders drainage. Another solution involves using a laser rangefinder to determine the height between the sensor and a specific component under the vehicle being inspected. The measured height during actual operation is compared to the height measured during deployment and teaching to obtain the height difference. Z-axis height compensation is then applied. However, currently available laser rangefinders have low measurement accuracy, resulting in missed or false detections. More importantly, they can only compensate for Z-axis height at a few inspection points, not all. Analysis of these methods reveals that existing approaches cannot simply and efficiently solve the problem of accurate train positioning by inspection robots. Summary of the Invention

[0006] This invention addresses the technical problems of existing inspection robot positioning and navigation methods, such as low compensation accuracy, susceptibility to external interference, long processing time, and low maintenance efficiency. It proposes a new comprehensive positioning and navigation method for urban rail vehicle inspection robots. This method not only features high positioning accuracy, high maintenance efficiency, strong anti-interference capabilities, and wide applicability, but also eliminates the need for resurfacing, thus reducing costs.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] The integrated positioning and navigation method for urban rail vehicle inspection robots includes the following steps:

[0009] A spatial coordinate system is established with the inspection robot as the origin, the robot's forward direction as the X-axis, the inner side of the trench track as the Y-axis, and the robot's height as the Z-axis.

[0010] Several reflective stickers were randomly pasted at different locations in the trench track, and a QR code was pasted on one side inside the trench track.

[0011] The inspection robot began to walk at the front end of the trench track according to the planned path in the map;

[0012] When the LiDAR of the inspection robot detects the reflective sticker, it compensates the X and Y axis coordinates of the inspection robot by comparing the actual position of the detected reflective sticker on the map with the current position of the inspection robot. After compensation, it continues to walk according to the planned path until the inspection robot has visited all inspection points.

[0013] When the positioning camera of the inspection robot detects a specific component under the train, the positioning camera feeds back the X and Y coordinates of the specific component under the train based on the 2D information provided by the image and further compensates the X and Y coordinates of the inspection robot to achieve the initial positioning of the inspection robot.

[0014] Based on the Z-value of a specific component under the train car fed back by the positioning camera, the Z-axis coordinate of the inspection robot is height-compensated, and then it continues to walk along the planned path after compensation.

[0015] When the robot reaches an inspection point, its QR code detection camera calculates the height difference between itself and the QR code. It then compares this height difference with the height difference calculated during the deployment and teaching of the current inspection point, and performs height compensation on the inspection robot in the Z-axis coordinate until the inspection robot has traversed all inspection points.

[0016] In one embodiment, the walking motion of the inspection robot is divided into the following two cases:

[0017] Deployment teaching: Deployment teaching is performed on the inspection robot. Based on the distribution of the vehicle parts, the path of the robotic arm is planned, and the inspection robot records the relevant parameters at different inspection points.

[0018] Task execution: The inspection robot executes the task according to the planned path. During the actual task execution, the inspection robot records relevant parameters at different inspection points.

[0019] The relevant parameters include the parameters of each axis of the robotic arm, the height of the lifting column, the height of the QR code, and the position parameters of the inspection robot.

[0020] In one embodiment, height compensation for the Z-axis coordinate of the inspection robot is divided into the following three cases: when the wheels are worn but the stopping position has not changed, when the wheels are not worn but the stopping position of the train has changed, and when the wheels are worn and the stopping position of the train has also changed.

[0021] For the three situations mentioned above, the specific parts under the train are marked as point A, the lidars at the front and rear ends of the inspection robot are marked as points B and D respectively, the installation position of the positioning camera is marked as point C, the center of the inspection robot is marked as point E, the center of the QR code detection camera is marked as point F, and the pasting position of the QR code is marked as point G.

[0022] Z a Z represents the diameter of the train wheels, and is a variable value; b This indicates the height of the end of the bollard from a specific component under the train; this value must be kept constant. c This represents the height of the positioning camera from a specific component under the train. The distance is calculated by taking pictures of the specific component with the positioning camera, and is a variable value; Z d This represents the height of the lifting column's end from the robot's center; it is a variable value. The lifting column is installed on the inspection robot, and its height is adjustable. e Z represents the height of the positioning camera from the center of the inspection robot, which is a fixed value; f This represents the height of the QR code detection camera from the center of the inspection robot, which is a fixed value; Z g This represents the height of the QR code detection camera relative to the QR code, and is a variable value; Z h This indicates the height of the QR code from a specific component under the train, and is a variable value; Z i This represents the height of the QR code from the track plane, and is a fixed value.

[0023] In one embodiment, when the wheels wear down but the parking position remains unchanged, the following method is used to compensate for the height of the inspection robot's Z-axis coordinate:

[0024] Due to height errors caused by wheel wear, during deployment and teaching, Z... b +Z d =Z c +Z e (1);

[0025] When the inspection robot performs its task according to the planned path, Z b +Z' d =Z' c +Z e (2), where Z e It is a fixed value, Z b These are values ​​that need to remain unchanged;

[0026] From the above formulas (1)-(2), we can derive Z' d =Z d +(Z' c -Z cThis allows us to determine the current height of the bollard, which is the height that the bollard needs to be adjusted to (Z). c -Z' c If the value is positive, the bollard needs to be lowered; if the value is negative, the bollard needs to be raised.

[0027] In one embodiment, when the wheels are not worn but the train's stopping position changes, the following method is used to compensate for the height of the inspection robot's Z-axis coordinate:

[0028] Due to the varying heights of the drainage ditches at different parking locations, Z... b =Z h -(Z) d +Z f +Z g (3);

[0029] When the inspection robot performs its task according to the planned path, Z b =Z h -(Z) d +Z f +Z" g (4), where Z b It needs to remain unchanged, Z f It is a fixed value because the wheel has not worn out, therefore Z is a fixed value in this case. h It is also a fixed value;

[0030] From the above formulas (3)-(4), we can derive Z". d =Z d +(Z g -Z" g This gives the current height of the bollard, which is the height that the bollard needs to be adjusted to (Z). g -Z" g If the value is positive, the bollard needs to be lowered; if the value is negative, the bollard needs to be raised.

[0031] In one embodiment, when wheel wear occurs and the train's stopping position changes, the following method is used to compensate for the height of the inspection robot's Z-axis coordinate:

[0032] The height error caused by wheel wear and the height error caused by the unevenness of the pit under different parking positions are superimposed;

[0033] From formulas (1)-(4), we can derive:

[0034] Given the current height of the bollard, the required height adjustment is [value missing]. If the value is positive, the bollard needs to be lowered; if the value is negative, the bollard needs to be raised.

[0035] In one embodiment, several reflective stickers are randomly pasted on the pillars on both sides of the trench, on the outside of both sides of the trench, at the end of the trench, and on the steps.

[0036] The inspection robot affixes QR codes to one side of the trench track within its operating range, maintaining the same spacing between adjacent QR codes. All QR codes are kept horizontal, parallel to the track plane above them, and installed within the field of view of the QR code detection camera. Each QR code contains unique height information.

[0037] In one embodiment, the selection principle for the specific components under the train is as follows: the component is fixed at the bottom of the vehicle, there is such component under each car, and its location is convenient for the positioning camera to take pictures, and components with obvious shape characteristics are preferred.

[0038] In one embodiment, the inspection point is the location where the inspection robot stops during the maintenance process.

[0039] In one embodiment, the inspection robot includes:

[0040] The robot itself,

[0041] The lifting column is fixed at one end to the top of the robot body, and its height is adjustable.

[0042] The robotic arm, a six-degree-of-freedom robotic arm, is mounted at the other end of the lifting column.

[0043] A positioning camera, mounted on top of the robot and in the middle area near the rear of the train, is used to photograph specific components under the train.

[0044] A QR code detection camera is installed on the outside of the robot body, near the center of the side, at the same height as the QR code. It is used to scan the QR code to obtain height information.

[0045] The lidar is fixedly installed at both the front and rear ends of the robot body to scan reflective stickers pasted at different locations on the trench track.

[0046] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0047] 1. The integrated positioning and navigation method for urban rail vehicle inspection robots proposed in this invention, without altering the original civil engineering of the trench tracks, compensates for height errors caused by unevenness in the trenches by combining the application of QR codes on one side of the inner wall of the trench with the calculation of distances from specific components under the vehicle by a positioning camera. This ensures the positioning accuracy of the inspection robot in the Z-axis direction and solves the problem that existing methods for compensating for unevenness in trench tracks require removing the original tiles inside the trench and relaid them to ensure flatness, which also delays normal maintenance operations and increases costs during the relaid process.

[0048] 2. The integrated positioning and navigation method for urban rail vehicle inspection robots proposed in this invention uses a 3D positioning camera for visual ranging, which has high measurement accuracy and strong anti-interference ability. It can perform overall compensation for height in the Z-axis direction, solving the problem that the existing common solution uses laser sensor ranging to compensate for errors, but the laser sensor ranging accuracy is low and it is easily affected by external interference.

[0049] 3. The integrated positioning and navigation method for urban rail vehicle inspection robots proposed in this invention improves the positioning accuracy in the X and Y axes by attaching reflective stickers on top of lidar navigation. The reflective stickers are relatively inexpensive. The QR code and QR code detection camera are used to compensate for the height error in the Z axis. These devices are also relatively inexpensive, thus reducing the overall cost of the positioning solution.

[0050] 4. The integrated positioning and navigation method for urban rail vehicle inspection robots proposed in this invention not only requires only one teaching demonstration for the same vehicle model, but can also be directly reused in other vehicles of the same model. Furthermore, it is highly valuable for the positioning of other vehicle models, and the entire positioning scheme has strong compatibility. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the trench track support column structure provided in an embodiment of the present invention;

[0052] Figure 2 This is a structural block diagram illustrating the source of positioning error in the inspection robot provided in an embodiment of the present invention.

[0053] Figure 3 This is a block diagram of the positioning error compensation structure of the inspection robot provided in an embodiment of the present invention;

[0054] Figure 4 This is a diagram showing the positional relationship between the inspection robot and the vehicle being inspected, provided in an embodiment of the present invention.

[0055] Figure 5 This is a simplified schematic diagram illustrating the positional relationship between the inspection robot and the vehicle being inspected, as provided in an embodiment of the present invention.

[0056] Figure 6 This is a flowchart of the X and Y axis direction compensation process for the inspection robot provided in an embodiment of the present invention;

[0057] Figure 7 This is a flowchart of the Z-axis direction compensation process for the inspection robot provided in an embodiment of the present invention;

[0058] Figure 8 This is a simplified diagram illustrating the positional relationship between the inspection robot and the vehicle being inspected when the wheels are worn but the parking position remains unchanged, provided as an embodiment of the present invention.

[0059] Figure 9 This is a simplified diagram illustrating the positional relationship between the inspection robot and the vehicle being inspected when the train's stopping position changes and the wheels are not worn, as provided in an embodiment of the present invention.

[0060] In the above figures:

[0061] 1. Underground trench track; 2. Inspection robot; 3. Column; 4. Underground trench track plan; 5. Track plan; 6. A train car; 7. QR code plan. Detailed Implementation

[0062] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] This invention provides a comprehensive positioning and navigation method for an urban rail vehicle inspection robot, comprising the following steps:

[0064] S1. Establish a spatial coordinate system with the inspection robot as the origin, the forward direction of the inspection robot as the X-axis, the inner side of the trench track as the Y-axis, and the height direction of the inspection robot as the Z-axis.

[0065] S2. Randomly affix several reflective stickers at different locations in the trench track, and affix a QR code on one side inside the trench track;

[0066] S3. The inspection robot begins to walk at the front end of the trench track according to the planned path in the map;

[0067] S4. When the LiDAR of the inspection robot detects the reflective sticker, the X and Y coordinates of the inspection robot are compensated by comparing the actual position of the detected reflective sticker on the map with the current position of the inspection robot. After compensation, the robot continues to walk along the planned path until it has completed all inspection points.

[0068] S5. When the positioning camera of the inspection robot detects a specific component under the train, the positioning camera feeds back the X and Y coordinates of the specific component under the train based on the 2D information provided by the image and further compensates the X and Y coordinates of the inspection robot to achieve the initial positioning of the inspection robot.

[0069] Regarding steps S4-S5 above, the X and Y axis positioning error compensation is explained as follows: Within the trench, the inspection robot moves to the center pin (a specific component under the vehicle). After the positioning camera captures the center pin, it calculates the deviation of the inspection robot relative to the center pin in the X and Y directions using the 2D information of the image. The inspection robot then compensates for these deviations, thus achieving initial positioning within the trench. The inspection robot uses laser SLAM navigation within the trench; however, the environment within the trench is too similar, hindering the robot's positioning and navigation. To assist laser SLAM navigation and improve positioning accuracy, reflective stickers are randomly affixed to the pillars on both sides of the trench and, in the last 10 meters or so of the trench, artificial features are created (i.e., artificially created protrusions on the side walls with randomly affixed reflective stickers, and randomly affixed reflective stickers to the steps behind). A map of the trench's tracks is pre-built, and feature points (reflective stickers) are recorded. These reflective stickers are positioned fixedly within the constructed map. As the inspection robot moves, the LiDAR scanner scans each reflective sticker, comparing the robot's current position and compensating for any deviations. This continuous positional compensation ensures the robot's accuracy. Actual testing showed that the inspection robot achieved a positioning and navigation accuracy of ±5mm in the X and Y directions throughout the entire trench.

[0070] S6. Based on the Z-value of the specific components under the train car fed back by the positioning camera, the Z-axis coordinate of the inspection robot is height-compensated, and then it continues to walk according to the planned path.

[0071] S7. When the robot reaches an inspection point, the robot's QR code detection camera calculates the height difference between itself and the QR code. The height difference is then compared with the height difference calculated during the deployment teaching of the current inspection point. The robot's height is compensated in the Z-axis coordinate until the robot has completed all inspection points.

[0072] In the above embodiments, the present invention provides a comprehensive positioning and navigation method for an urban rail vehicle inspection robot. The purpose of this method is to address the problems of existing positioning and navigation methods for inspection robots, such as low compensation accuracy, susceptibility to external interference, long processing time, and low maintenance efficiency. Before solving these problems, it is necessary to first describe the walking motion of the inspection robot, which can be divided into the following two situations:

[0073] (1) Deployment teaching: Deployment teaching is carried out on the inspection robot. Based on the distribution of the vehicle body parts, the path of the robotic arm is planned. The inspection robot records the relevant parameters at different inspection points (the relevant parameters include the parameters of each axis of the robotic arm, the height of the lifting column, the height of the QR code, and the position parameters of the inspection robot).

[0074] (2) Task execution: The inspection robot executes the task according to the planned path. During the actual task execution, the inspection robot records relevant parameters at different inspection points (relevant parameters include the parameters of each axis of the robotic arm, the height of the lifting column, the height of the QR code, and the position parameters of the inspection robot). However, during the task execution, it will be affected by factors such as different train stopping positions and wheel wear, which will cause the relevant parameters measured during the task execution to be different from those during the deployment and teaching. This is also the reason why the inspection robot is currently unable to accurately locate the train.

[0075] Furthermore, to address the aforementioned problems, this invention uses the inspection robot as the coordinate origin, sets the robot's forward direction as the X-axis, the direction inward of the trench track as the Y-axis, and the robot's height direction as the Z-axis, analyzing these aspects in two ways: X and Y axes, and Z axis. Specifically:

[0076] Regarding the X and Y axes: Since most of the sections along the trench are made of identical pillars, and the last 10 meters or so are entirely walls with almost no distinguishing features, the inspection robot's working environment is highly similar, preventing it from achieving high-precision trackless navigation and positioning. If the inspection robot cannot achieve accurate positioning in the X and Y directions, the posture of the vehicle components photographed during inspection will differ significantly from the posture photographed during deployment and teaching, potentially even causing components to be outside the camera's field of view. This prevents the visual algorithm from effectively processing the image. A common solution is to increase the camera's field of view to ensure vehicle components appear as centrally as possible, while also improving the compatibility of the visual algorithm. However, this method cannot guarantee that all vehicle components to be inspected will be within the camera's field of view.

[0077] Regarding the Z-axis direction: The design characteristics of the subway maintenance depot's trench tracks are such that, to facilitate drainage, the tracks are uneven and undulating. This results in the trench track surfaces not being on the same horizontal plane; the maximum height difference measured along the entire trench track can reach 10cm. After subway vehicles have been running on the mainline for a period of time, the wheels will wear down to some extent, causing the overall height of the train to decrease. The maximum height difference between new and old trains can reach 2-3cm. Furthermore, maintenance work on the train can also cause changes in its height. These factors all contribute to the positioning error of the inspection robot along the Z-axis. Common solutions to compensate for this error include repaving the trench track surface or using a laser rangefinder to obtain the height between the sensor and a specific component under the vehicle being inspected. The height value measured during actual operation is compared with the height value measured during deployment and teaching to obtain the height difference. These methods not only consume a lot of manpower and resources, but the laser rangefinder's measurement accuracy is also relatively low, failing to fundamentally solve the problem of Z-axis positioning error compensation. Therefore, this invention proposes a novel integrated positioning and navigation method for urban rail vehicle inspection robots. This method improves the positioning accuracy in the X and Y axes by attaching reflective stickers on top of lidar navigation, compensates for height errors in the Z axis by using QR codes and QR code detection cameras, and performs visual ranging using a 3D positioning camera. This method has high measurement accuracy, strong anti-interference capabilities, and can perform overall height compensation in the Z axis direction.

[0078] Furthermore, the working environment and component composition of the inspection robot of this invention are described below: The ground of the track trench is uneven and not on the same horizontal plane, and the train wheels will also wear. The train stops at different positions each time, with a maximum error of no more than 1 meter. Each train to be inspected has two bogies, and each bogie has a center pin (i.e., a specific component under the car body) located at the center of the bogie. The lifting column is mounted on the robot, with a fixed installation position, but the height of the lifting column is adjustable. At the end of the lifting column is a six-degree-of-freedom robotic arm, and a positioning camera is mounted at the end of the robotic arm. The positioning camera is mounted on the inspection robot, with a fixed installation position. The positioning camera takes pictures of the center pin under the car body and calculates the offset of the inspection robot relative to the center pin in the X, Y, and Z axis directions. The position where the inspection robot stops during the maintenance process is called an "inspection point." At each inspection point, the camera at the end of the robotic arm can take pictures and identify several car body components. The QR code detection camera is installed on the inspection robot in a fixed position. At each inspection point, the height difference between the QR code camera and the QR code can be calculated. The QR code is affixed to the wall on one side of the trench. A lidar is installed at the front and rear of the robot. The lidar can scan the reflective sticker.

[0079] In one specific implementation, height compensation for the Z-axis coordinate of the inspection robot is divided into the following three cases: when the wheels are worn but the stopping position has not changed; when the wheels are not worn but the stopping position of the train has changed; and when the wheels are worn and the stopping position of the train has also changed.

[0080] To compensate for the height error caused by the unevenness of the trench at different parking positions, this invention uses a 250-meter-long trench track as an example. This trench track has six roughly evenly distributed drainage troughs, which are the lowest points of the trench plane. Each drainage trough roughly corresponds to one train car. The inspection robot is 909cm long, which is very small compared to the entire length of the trench. Therefore, when the inspection robot is in the trench area, the change in the height of the trench plane in the adjacent area can be simplified as follows: Figure 5 The elevation changes of the trench plane shown indicate that the trench track plane can be simplified into sections of planes with different heights.

[0081] For the three situations mentioned above, the specific parts under the train are marked as point A, the lidars at the front and rear ends of the inspection robot are marked as points B and D respectively, the installation position of the positioning camera is marked as point C, the center of the inspection robot is marked as point E, the center of the QR code detection camera is marked as point F, and the pasting position of the QR code is marked as point G.

[0082] Z a Z represents the diameter of the train wheels, and is a variable value; b This indicates the height of the end of the bollard from a specific component under the train; this value must be kept constant. c This represents the height of the positioning camera from a specific component under the train. The distance is calculated by taking pictures of the specific component with the positioning camera, and is a variable value; Z d This represents the height of the lifting column's end from the robot's center; it is a variable value. The lifting column is installed on the inspection robot, and its height is adjustable. e Z represents the height of the positioning camera from the center of the inspection robot, which is a fixed value; f This represents the height of the QR code detection camera from the center of the inspection robot, which is a fixed value; Z g This represents the height of the QR code detection camera relative to the QR code, and is a variable value; Z h This indicates the height of the QR code from a specific component under the train, and is a variable value; Z i This represents the height of the QR code from the track plane, and is a fixed value.

[0083] In one specific implementation, when the wheels wear down but the parking position remains unchanged (as shown in the diagram of the positional relationship between the inspection robot and the vehicle being inspected), Figure 8As shown, the following method is used to compensate for the height of the inspection robot's Z-axis coordinate:

[0084] Due to height errors caused by wheel wear, during deployment and teaching, Z... b +Z d =Z c +Z e (1);

[0085] When the inspection robot performs its task according to the planned path, Z b +Z' d =Z' c +Z e (2), where Z e It is a fixed value, Z b These are values ​​that need to remain unchanged;

[0086] From the above formulas (1)-(2), we can derive Z' d =Z d +(Z' c -Z c This allows us to determine the current height of the bollard, which is the height that the bollard needs to be adjusted to (Z). c -Z' c If the value is positive, the bollard needs to be lowered; if the value is negative, the bollard needs to be raised.

[0087] In one specific implementation, when the wheels are not worn, but the train's stopping position changes (as shown in the diagram of the positional relationship between the inspection robot and the inspected vehicle), Figure 9 As shown, the following method is used to compensate for the height of the inspection robot's Z-axis coordinate:

[0088] Due to the varying heights of the drainage ditches at different parking locations, Z... b =Z h -(Z) d +Z f +Z g (3);

[0089] When the inspection robot performs its task according to the planned path, Z b =Z h -(Z) d +Z f +Z" g (4), where Z b It needs to remain unchanged, Z f It is a fixed value because the wheel has not worn out, therefore Z is a fixed value in this case. h It is also a fixed value;

[0090] From the above formulas (3)-(4), we can derive Z".d =Z d +(Z g -Z" g This gives the current height of the bollard, which is the height that the bollard needs to be adjusted to (Z). g -Z" g If the value is positive, the bollard needs to be lowered; if the value is negative, the bollard needs to be raised.

[0091] In one specific implementation, when the wheels wear out and the train's stopping position changes, the following method is used to compensate for the height of the inspection robot's Z-axis coordinate:

[0092] The height error caused by wheel wear and the height error caused by the unevenness of the pit under different parking positions are superimposed;

[0093] From formulas (1)-(4), we can derive:

[0094] Given the current height of the bollard, the required height adjustment is [value missing]. If the value is positive, the bollard needs to be lowered; if the value is negative, the bollard needs to be raised.

[0095] In one specific implementation, several reflective stickers are randomly pasted on the pillars on both sides of the trench, the exterior of both sides of the trench, the end of the trench, and the steps.

[0096] The inspection robot affixes QR codes to one side of the trench track within its operating range, maintaining the same spacing between adjacent QR codes. All QR codes are kept horizontal, parallel to the track plane above them, and installed within the field of view of the QR code detection camera. Each QR code contains unique height information.

[0097] In one specific embodiment, the selection principle for the specific components of the train undercarriage is as follows: the component is fixed at the bottom of the vehicle, there is such component under each car, and its location is convenient for the positioning camera to take pictures. Components with obvious shape characteristics are preferred, such as center pins.

[0098] In one specific embodiment, the inspection robot includes:

[0099] The robot itself,

[0100] The lifting column is fixed at one end to the top of the robot body, and its height is adjustable.

[0101] The robotic arm, a six-degree-of-freedom robotic arm, is mounted at the other end of the lifting column.

[0102] A positioning camera, mounted on top of the robot and in the middle area near the rear of the train, is used to photograph specific components under the train.

[0103] A QR code detection camera is installed on the outside of the robot body, near the center of the side, at the same height as the QR code. It is used to scan the QR code to obtain height information.

[0104] The lidar is fixedly installed at both the front and rear ends of the robot body to scan reflective stickers pasted at different locations on the trench track.

[0105] To provide a clearer and more detailed description of the integrated positioning and navigation method for urban rail vehicle inspection robots provided in this invention, specific embodiments will be described below.

[0106] Example 1

[0107] This embodiment provides a comprehensive positioning and navigation method for an urban rail vehicle inspection robot, specifically including a positioning error compensation method for the inspection robot in the X and Y axes, as follows:

[0108] (1) As Figure 6 As shown in the flowchart, the inspection robot starts walking at the front end of the trench. If no reflective sticker is detected, it follows the planned path on the map. If the lidar detects a reflective sticker, it compares the actual position of the reflective sticker on the map with the robot's current position, and then compensates the robot's X and Y axis coordinates in the navigation algorithm. After X and Y axis coordinate compensation, the inspection robot continues to walk along the path on the map.

[0109] (2) When the positioning camera detects the center pin under the vehicle, it will compensate the X and Y coordinates of the robot based on the 2D information of the image and the X and Y values ​​fed back by the positioning camera to achieve the initial positioning of the robot. After the X and Y coordinates are compensated, the inspection robot will continue to walk along the path planned by the map. That is, as long as the LiDAR detects the reflective sticker during the walking process, the X and Y coordinates of the inspection robot will be compensated in the navigation algorithm until the inspection robot has walked through all the inspection points.

[0110] Example 2

[0111] This embodiment provides a comprehensive positioning and navigation method for an urban rail vehicle inspection robot, specifically including a method for compensating for positioning errors in the Z-axis direction of the inspection robot, as follows:

[0112] (1) As Figure 7As shown in the flowchart, following the path planned on the map, the inspection robot begins to walk at the front end of the trench. If the positioning camera does not detect the center pin, the inspection robot will continue to walk along the planned path. When the positioning camera detects the center pin, it will compensate for the Z-axis direction error based on the Z value fed back by the positioning camera. This is mainly to compensate for the error caused by wheel wear.

[0113] (2) After Z-axis compensation, the inspection robot will continue to walk along the path planned on the map. When it reaches an inspection point, the QR code camera will calculate the height difference between itself and the QR code. This height difference will be compared with the height difference calculated during the teaching process for the current inspection point, and then the Z-axis height will be compensated. This is mainly to compensate for the error caused by the unevenness of the ground surface of the trench track. That is, the Z-axis height difference will be compensated at each inspection point until the maintenance robot has walked through all the inspection points.

Claims

1. A comprehensive positioning and navigation method for urban rail vehicle inspection robots, characterized in that, Includes the following steps: A spatial coordinate system is established with the inspection robot as the origin, the robot's forward direction as the X-axis, the inner side of the trench track as the Y-axis, and the robot's height as the Z-axis. Several reflective stickers were randomly pasted at different locations in the trench track, and a QR code was pasted on one side inside the trench track. The inspection robot began to walk at the front end of the trench track according to the planned path in the map; When the lidar on the inspection robot detects the reflective sticker, the robot's X and Y axis coordinates are compensated by comparing the actual position of the detected reflective sticker on the map with the current position of the inspection robot. After compensation, the robot continues to walk along the planned path until it has visited all inspection points. When the positioning camera of the inspection robot detects a specific component under the train, the positioning camera feeds back the X and Y coordinates of the specific component under the train based on the 2D information provided by the image and further compensates the X and Y coordinates of the inspection robot to achieve the initial positioning of the inspection robot. The selection principle for the specific component under the train is: the component is fixed at the bottom of the vehicle, there is such component under each car, and its location is convenient for the positioning camera to take pictures. The inspection robot performs height compensation on the Z-axis coordinate based on the Z-value of the specific component under the train as fed back by the positioning camera, and continues to walk along the planned path after compensation. When the robot reaches an inspection point, its QR code detection camera calculates the height difference between itself and the QR code. Then, it compares the height difference with the height difference calculated during the deployment teaching of the current inspection point and performs height compensation on the inspection robot in the Z-axis coordinate until the inspection robot has completed all inspection points. The walking motion of the inspection robot can be divided into the following two cases: Deployment teaching: Deployment teaching is performed on the inspection robot. Based on the distribution of the vehicle parts, the path of the robotic arm is planned, and the inspection robot records the relevant parameters at different inspection points. Task execution: The inspection robot executes the task according to the planned path. During the actual task execution, the inspection robot records relevant parameters at different inspection points. The relevant parameters include the parameters of each axis of the robotic arm, the height of the lifting column, the height of the QR code, and the position parameters of the inspection robot. Height compensation for the Z-axis coordinate of the inspection robot is divided into the following three cases: when the wheels are worn but the stopping position has not changed; when the wheels are not worn but the stopping position of the train has changed; and when the wheels are worn and the stopping position of the train has also changed. For the three situations mentioned above, the specific component under the train is marked as point A, the lidars at the front and rear ends of the inspection robot are marked as points B and D respectively, the installation position of the positioning camera is marked as point C, the center of the inspection robot is marked as point E, the center of the QR code detection camera is marked as point F, and the pasting position of the QR code is marked as point G. Za represents the diameter of the train wheel, a variable value; Zb represents the height of the end of the lifting column from a specific component under the train, which must be kept constant; Zc represents the height of the positioning camera from the specific component under the train, calculated by taking pictures of the component under the train using the positioning camera, a variable value; Zd represents the height of the end of the lifting column from the center of the robot, a variable value; the lifting column is installed on the inspection robot, and its height can be adjusted; Ze represents the height of the positioning camera from the center of the inspection robot, a fixed value; Zf represents the height of the QR code detection camera from the center of the inspection robot, a fixed value; Zg represents the height of the QR code detection camera from the QR code, a variable value; Zh represents the height of the QR code from a specific component under the train, a variable value; Zi represents the height of the QR code from the track plane, a fixed value. When the wheels are not worn, but the train's stopping position changes, the following method is used to compensate for the height of the inspection robot's Z-axis coordinate: Due to the height error caused by the undulation of the underground ditch under different parking positions, when deploying the teaching, Zb=Zh-(Zd+Zf+Zg)(3); When the inspection robot performs its task according to the planned path, Zb = Zh - (Z"d + Zf + Z"g) (4), where Zb needs to remain unchanged, Zf is a fixed value, and Zh is also a fixed value in this case because the wheels have not worn. From the above formulas (3)-(4), we can obtain Z"d=Zd+(Zg-Z"g), which gives the height value of the lifting column under the current situation. That is, the height value that the lifting column needs to be adjusted is (Zg-Z"g). If the value is positive, the lifting column needs to be lowered. If the value is negative, the lifting column needs to be raised.

2. The integrated positioning and navigation method for urban rail vehicle inspection robots according to claim 1, characterized in that, When the wheels wear out but the parking position remains unchanged, the following method is used to compensate for the height of the inspection robot's Z-axis coordinate: Due to the height error caused by wheel wear, when deploying the teaching, Zb + Zd = Zc + Ze (1); When the inspection robot performs its task according to the planned path, Zb + Z'd = Z'c + Ze (2), where Ze is a fixed value and Zb is a value that needs to remain unchanged; From the above formulas (1)-(2), we can obtain Z'd=Zd+(Z'c-Zc), which gives the height value of the lifting column under the current situation. That is, the height value that the lifting column needs to be adjusted is (Zc-Z'c). If the value is positive, the lifting column needs to be lowered. If the value is negative, the lifting column needs to be raised.

3. The integrated positioning and navigation method for urban rail vehicle inspection robots according to claim 2, characterized in that, When the wheels wear out and the train's stopping position changes, the following method is used to compensate for the height of the inspection robot's Z-axis coordinate: The height error caused by wheel wear and the height error caused by the unevenness of the pit under different parking positions are superimposed; From formulas (1)-(4), we can derive: Z"d = [2Zd + (Zg - Z"g) + (Z'c - Zc)] / 2, which gives the current height of the bollard. The bollard needs to be adjusted to a height of [(Zg - Z"g) + (Z'c - Zc)] / 2. If the value is positive, the bollard needs to be lowered; if the value is negative, the bollard needs to be raised.

4. The integrated positioning and navigation method for the urban rail vehicle inspection robot according to claim 1, characterized in that, Reflective stickers were randomly pasted on the pillars on both sides of the trench, on the outside of both sides of the trench, at the end of the trench, and on the steps. The inspection robot affixes QR codes to one side of the trench track within its operating range, maintaining the same spacing between adjacent QR codes. All QR codes are kept horizontal, parallel to the track plane above them, and installed within the field of view of the QR code detection camera. Each QR code contains unique height information.

5. The integrated positioning and navigation method for the urban rail vehicle inspection robot according to claim 1, characterized in that, The inspection point is the location where the inspection robot stops during the maintenance process.

6. The integrated positioning and navigation method for the urban rail vehicle inspection robot according to claim 1, characterized in that, The inspection robot includes: The robot itself; The lifting column is fixed at one end to the top of the robot body and its height is adjustable. The robotic arm, which is a six-degree-of-freedom robotic arm, is installed at the other end of the lifting column. A positioning camera is installed on top of the robot body, in the middle area near the rear of the vehicle, to take pictures of specific components under the train. A QR code detection camera is installed on the outside of the robot body, in the area near the center of the side of the vehicle. Its installation height is consistent with the height of the QR code, and it is used to scan the QR code to obtain height information. The lidar is fixedly installed at both the front and rear ends of the robot body to scan reflective stickers pasted at different locations on the trench track.

Images