Point cloud real-time mapping and positioning system and method for dual-axle steering unmanned mining trucks

Abstract

This invention relates to the field of unmanned driving environment perception technology. It proposes a real-time point cloud mapping and positioning system and method for dual-axle steering unmanned mining trucks. This system achieves real-time, accurate mapping by registering and correcting odometer information with forward and backward laser radars. Furthermore, the forward and backward laser radars are matched with a priori maps to perform coarse and fine positioning, respectively, achieving precise vehicle positioning. This system enhances the environmental perception capabilities of dual-axle steering mining trucks, enables precise positioning to meet operational requirements, ensures worker safety, and improves production efficiency in mining areas.

Description

Technical Field

[0001] The present invention relates to the field of unmanned driving environment perception technology, and in particular to a point cloud real-time mapping and positioning system and method for dual-axle steering unmanned mining trucks. Background Art

[0002] With the rapid development of 5G communications and connected vehicle technologies, autonomous driving has become a research hotspot. However, due to issues such as sensor stability, decision reliability, and control accuracy, autonomous driving is unlikely to be widely adopted in urban areas in the short term. However, open-pit mining areas offer the natural advantages of fewer people and fewer obstacles, making autonomous driving applications more suitable.

[0003] As a new type of cab-less vehicle, dual-axle steering unmanned mining trucks offer long operating times, low labor consumption, and high flexibility, making them well-suited to mining environments. However, they fundamentally require highly reliable environmental perception technology; otherwise, they cannot accurately locate and provide obstacle avoidance warnings. As a key environmental perception technology, high-precision maps not only provide information about surrounding obstacles and road conditions, but also provide precise positioning information based on the vehicle's surroundings, safeguarding unmanned mining trucks. Currently, high-precision mapping in mining areas often relies on a brute-force approach, stacking LiDAR point clouds frame by frame. Camera image data is then mapped onto the LiDAR point cloud. However, this approach consumes a lot of memory and takes a long time to create maps. It also lacks real-time mapping capabilities and can significantly deviate from the actual scene after completion. This makes it unable to cope with sudden environmental changes, thus reducing positioning accuracy. Furthermore, during the positioning process, the initial pose provided by a one-way vehicle during return is more susceptible to deviations due to factors such as reversing. Furthermore, a single LiDAR cannot correct the positioning results, making it difficult to accurately locate the vehicle. Therefore, how to use lidar to map the mining environment in real time and determine the vehicle positioning is an urgent problem to be solved.

[0004] Currently, lidar mapping and positioning have become a research hotspot, but they are difficult to meet the application requirements in this scenario:

[0005] Chinese patent publication number CN201911002173.6, titled "A LiDAR Mapping Method, Apparatus, Equipment, and Medium for Indoor Scenarios," primarily provides a method for mapping indoors using LiDAR integrated with UWB. This method uses UWB and a laser odometry to correct vehicle position for positioning. However, the invention has not been applied outdoors, and the UWB positioning network in mining areas is imperfect, making positioning accuracy difficult to guarantee.

[0006] Chinese patent publication number CN202010561444.8, titled "Lidar Positioning Method and Related Device," provides a positioning method that performs multiple point cloud matching and screening based on a predicted initial pose, enabling relatively accurate vehicle positioning. However, the invention does not address the generation method and process of a grid map, and is merely a theoretical method hypothesis, without any application scenario construction.

[0007] In the paper "Research on AGV Navigation Based on LiDAR Positioning and Visual Redundancy Confirmation, Chen Cheng, Shanghai University, 2021. DOI: 10.27300 / d.cnki.gshau.2021.000274," a positioning method combining LiDAR and visual redundancy is proposed. The automated navigation vehicle uses LiDAR to scan wall reflectors and visually scans artificial landmarks for positioning confirmation, continuously correcting the positioning process to achieve precise positioning and, to a certain extent, eliminating cumulative errors. However, this method has limited application in mining scenarios. The distance between working areas in mining areas makes it impossible to eliminate cumulative errors in a timely manner, resulting in inaccurate LiDAR positioning. Summary of the Invention

[0008] To address the aforementioned challenges in the existing technology, the present invention proposes a real-time point cloud mapping and positioning system and method for dual-axle steering unmanned mining trucks. This system achieves real-time, accurate mapping by registering and correcting odometry information with forward and backward LiDARs. Furthermore, the forward and backward LiDARs are matched with a priori maps to perform coarse and fine positioning, respectively, achieving precise positioning of the vehicle. This enhances the environmental perception capabilities of dual-axle steering mining trucks, enabling precise positioning to meet operational requirements, ensuring worker safety, and improving production efficiency in mining areas.

[0009] The technical solutions of the present invention are as follows:

[0010] A laser radar real-time mapping and positioning system for a dual-axle steering unmanned mining truck, comprising two laser radars respectively arranged at the front and rear of the dual-axle steering unmanned mining truck, a laser radar calibration module, a path planning module, a laser radar mapping module, and a laser radar positioning module;

[0011] The laser radar calibration module performs parameter calibration based on the intrinsic and extrinsic parameters of the two laser radars through normal distribution point cloud registration, unifying the two laser radars in the vehicle body coordinate system;

[0012] The path planning module uses the Dijkstra algorithm to select the shortest path as the optimal path in the area of ​​interest to the user of the dual-axle steering unmanned mining truck, and plans the vehicle driving state based on the optimal path using the simulated annealing algorithm;

[0013] The LiDAR mapping module generates laser odometry information based on the forward LiDAR, providing the vehicle's relative position and posture information. It generates a map based on point cloud feature information matching based on the backward LiDAR, performs loop correction optimization on the laser odometry information of the forward LiDAR, and performs point cloud degrounding, screening, clustering, and matching on the generated map to ultimately form a three-dimensional map. The measurement and calculation results of the two LiDARs are uploaded to the cloud, which matches the measurement and calculation results of the two according to the time frame to form a complete map.

[0014] The LiDAR positioning module uses the LiDAR point cloud feature information to match the normally distributed point cloud with the complete map in the cloud, generates probability information, and determines the positioning position of the LiDAR.

[0015] Preferably, the front and rear laser radars are respectively placed above the front and rear bumpers of the unmanned mining truck.

[0016] The present invention also provides a real-time laser radar mapping method for a dual-axle steering unmanned mining truck, comprising the following steps:

[0017] S1 users calibrate the parameters of the front and rear LiDARs of a dual-axle steering unmanned mining truck based on their internal and external parameters and positional relationships.

[0018] S2 user selects an area of ​​interest from the map and selects a path based on the area of ​​interest as the planned path;

[0019] S3 activates the dual-axle steering unmanned mining truck and, after reaching the designated starting point of the planned path, turns on the front and rear lidars;

[0020] The S4 dual-axle steering unmanned mining truck travels along the planned route. During driving, the forward LiDAR A generates laser odometry information, and the rear LiDAR B collects map point cloud information. The laser odometry information of the forward LiDAR A is subjected to loopback correction. A map is then constructed based on the point cloud features in the point cloud information. The measurement and calculation results of the two LiDARs are fused to generate a complete point cloud map, which is then uploaded to the cloud.

[0021] Preferably, in step S1, the user inputs the measured extrinsic parameter value, and the computer performs iterative optimization of normal distribution feature matching based on the extrinsic parameter value and the point cloud generated by the two laser radars in a stationary state, so that the point clouds of the two laser radars can eventually reach a completely matched state.

[0022] Preferably, the planned path in step S2 is obtained using a shortest path algorithm.

[0023] Preferably, the laser odometer information in step S4 is generated by point cloud curvature feature information collected by a forward laser radar.

[0024] The present invention also provides a laser radar positioning method for a dual-axle steering unmanned mining truck, comprising the following steps:

[0025] After the S5 dual-axle steering unmanned mining truck reaches the end of the planned route, it switches the unmanned mining truck to the forward and backward directions and returns along the original planned route;

[0026] The S6's forward-facing lidar A and rearward-facing lidar B retrieve the cloud-based point cloud map, match the point clouds according to the normal distribution changes, and generate solution values. The solution values ​​are then transformed relative to the center point to generate positioning A and positioning B, respectively. The final vehicle positioning is generated based on weighted calculation.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] 1. This invention proposes a real-time LiDAR mapping and positioning system and method for dual-axle steering unmanned mining trucks. Leveraging the two-way flexibility of dual-axle steering trucks and the comparable sensing characteristics of forward and backward sensors, this system uses a forward LiDAR to provide odometer information, while a backward LiDAR performs loop correction on the forward odometer information through point cloud map registration. Combined with the SLAM real-time mapping method, this system utilizes less memory, offers improved real-time performance, and produces more accurate map output compared to brute-force mapping methods.

[0029] 2. The present invention proposes a real-time LiDAR mapping and positioning system and method for dual-axle steering unmanned mining trucks, which has the advantage of precise vehicle positioning based on forward and backward LiDAR matching and positioning. By utilizing the bidirectional driving characteristics of dual-axle steering unmanned mining trucks, reverse driving without position deviation can be achieved, perfectly solving the initial positioning pose problem. Combined with the forward and backward LiDAR point cloud matching and positioning results, weighted processing is performed, coarse positioning determines the approximate position, and fine positioning further determines the specific position. This allows for more precise vehicle positioning, with smaller sensor errors than single LiDAR positioning.

[0030] 3. Based on mining scenarios, this invention adopts multi-lidar real-time fusion mapping and positioning technology. The developed system is applied to dual-axle steering unmanned mining trucks. It has the characteristics of low memory consumption, high real-time performance, and high precision. It solves the problem of vehicle positioning in mining areas when GPS is lost, and forms an innovative perception method library for real-time mapping and positioning in mining environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. By referring to the drawings, the features and advantages of the present invention will be more clearly understood. The drawings are schematic and should not be understood as limiting the present invention in any way. Those skilled in the art can derive other drawings based on these drawings without inventive effort. Among them:

[0032] Figure 1 This is a flow chart of the real-time mapping and positioning method of the laser radar for dual-axle steering unmanned mining trucks of the present invention;

[0033] Figure 2 This is the real-time mapping effect diagram of the laser radar for the double-bridge steering unmanned mining truck of the present invention;

[0034] Figure 3 This is a rendering of the real-time positioning effect of the laser radar for the double-axle steering unmanned mining truck of the present invention. DETAILED DESCRIPTION

[0035] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present invention and the features therein can be combined with each other without conflict.

[0036] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0037] On one hand, the present invention proposes a real-time laser radar mapping and positioning system for a dual-bridge steering unmanned mining truck. The system includes two laser radars respectively arranged at the front and rear of the dual-bridge steering unmanned mining truck, a laser radar calibration module, a path planning module, a laser radar mapping module, and a laser radar positioning module.

[0038] The laser radar calibration module performs parameter calibration based on the intrinsic and extrinsic parameters of the two laser radars through normal distribution point cloud registration, unifying the two laser radars in the vehicle body coordinate system;

[0039] The path planning module uses the Dijkstra algorithm to select the shortest path as the optimal path in the area of ​​interest to the user of the dual-axle steering unmanned mining truck, and plans the vehicle driving state based on the optimal path using the simulated annealing algorithm;

[0040] The LiDAR mapping module generates laser odometry information based on the forward LiDAR, providing the vehicle's relative position and posture information. It generates a map based on point cloud feature information matching based on the backward LiDAR, performs loop correction optimization on the laser odometry information of the forward LiDAR, and performs point cloud degrounding, screening, clustering, and matching on the generated map to ultimately form a three-dimensional map. The measurement and calculation results of the two LiDARs are uploaded to the cloud, which matches the measurement and calculation results of the two according to the time frame to form a complete map.

[0041] The LiDAR positioning module uses the LiDAR point cloud feature information to match the normally distributed point cloud with the complete map in the cloud, generates probability information, and determines the positioning position of the LiDAR.

[0042] The present invention adopts a double-axle steering unmanned mining truck, which is a new type of cab-less vehicle with high maneuverability. LiDAR A and LiDAR B are respectively placed above the front and rear bumpers.

[0043] Based on the above system, the present invention further proposes a real-time mapping and positioning method of laser radar for dual-bridge steering unmanned mining trucks, the specific steps of which are as follows: Figure 1 Shown, including:

[0044] The user activates the dual-axle steering system on the unmanned mining truck, and the LiDAR begins operating. Using a ruler, the user measures the positional relationship between LiDARs A and B. The user then inputs the extrinsic parameters of the two LiDARs and performs feature matching on a normally distributed point cloud. This generates and records the precise rotation and translation matrices, which are then updated. This completes the LiDAR calibration process.

[0045] After S2 calibration is complete, the user selects a map area of ​​interest in the remote control platform box. A feasible path is found within the selected map area, and the shortest path from the Dijkstra algorithm is used as the planned path. Finally, a simulated annealing algorithm is used to obtain a smooth planned path. Based on the current and starting positions, a feasible section is found and the vehicle is driven to the starting position, achieving autonomous control of the dual-axle steering unmanned mining truck.

[0046] After the S3 dual-axle steering unmanned mining truck arrives at the starting point, it builds a map based on the data collected by the two lidars. LiDAR mapping is a composite process of generating odometers and building a three-dimensional map based on the features of the lidar point cloud. The odometer information provides location coordinate information, and the three-dimensional point cloud provides spatial shape information. Specifically,

[0047] The lightweight, real-time laser mapping solution called lego-loam is not only suitable for most scenarios, but also produces more detailed maps. The point cloud information collected by LiDAR A (forward) generates laser odometry information based on the curvature characteristics of the point cloud. The laser odometry can serve as an effective sensor for relative positioning of the vehicle, based on the movement between the point cloud data of the previous and next frames. The point cloud collected by LiDAR B (backward) is matched based on the feature information of the point cloud to generate a three-dimensional point cloud map, and the forward odometry information is loop-corrected. After the two LiDARs complete their work, they upload the results in real time to the cloud, which fuses the two to form a complete point cloud map with relative position information, and uploads the point cloud map in real time.

[0048] After the S4 point cloud map is established, the dual-bridge steering unmanned mining truck is controlled to switch modes, changing the original forward direction to backward and the backward direction to forward. That is, LiDAR B is switched to the current forward radar and LiDAR A is switched to the current backward radar. At the same time, the planned path is reversed. At this point, the dual-bridge steering unmanned mining truck mode switching is completed.

[0049] After the S5 dual-axle steering unmanned mining truck turns, the data collected by the lidar is redistributed to the execution task, and the lidar positioning is performed. The lidar point cloud information is used to match the normal distribution point cloud with the prior map to generate probability information, and the positioning position of the lidar is determined based on this. Specifically,

[0050] A complete point cloud map is obtained from the cloud as a prior map. LiDAR A performs point cloud normal distribution feature matching according to the prior map to obtain the current coarse location A of the vehicle. Based on the information of coarse location A, LiDAR B performs point cloud feature matching to output a fine location B.

[0051] S6 finally fuses the coarse positioning A and the fine positioning B. This method uses a weighted fusion approach, with fine positioning receiving a higher weight and coarse positioning receiving a lower weight. This fused positioning information is then uploaded to the cloud. This system completes its execution, and the dual-axle steering unmanned mining truck returns to the workshop.

[0052] Furthermore, the foregoing describes only some embodiments, which may be changed, modified, added and / or varied without departing from the scope and spirit of the disclosed embodiments, which are illustrative and not restrictive. Furthermore, the embodiments described relate to what are currently considered to be the most practical and preferred embodiments, and it should be understood that the embodiments should not be limited to the disclosed embodiments, but rather are intended to cover different modifications and equivalent arrangements that are included within the spirit and scope of the embodiments. Furthermore, the various embodiments described above may be used in conjunction with other embodiments, such as aspects of one embodiment may be combined with aspects of another embodiment to achieve yet another embodiment. Additionally, each independent feature or component of any given component may constitute another embodiment.

[0053] The foregoing description of the embodiments is provided for the purpose of illustration and description, and is not intended to be exhaustive or limit the present disclosure. Each element or feature of a specific embodiment is generally not limited to that specific embodiment, but in applicable cases, even if not specifically shown or described, each element or feature is also interchangeable and can be used for the embodiment of selection, and can also be changed in many ways. This change is not considered to be a deviation from the present disclosure, and all such changes are included within the scope of the present disclosure.

[0054] Therefore, it should be understood that the drawings and description herein are provided by way of illustration to facilitate understanding of the present invention and should not be construed as limiting the scope thereof.

Claims

1. A real-time laser radar mapping and positioning system for dual-axle steering unmanned mining trucks, characterized by: It includes two laser radars, a laser radar calibration module, a path planning module, a laser radar mapping module and a laser radar positioning module, which are respectively arranged at the front and rear of the dual-axle steering unmanned mining truck; the two laser radars are respectively placed above the front and rear bumpers of the unmanned mining truck; The laser radar calibration module performs parameter calibration based on the intrinsic and extrinsic parameters of the two laser radars through normal distribution point cloud registration, unifying the two laser radars in the vehicle body coordinate system; The path planning module uses the Dijkstra algorithm to select the shortest path as the optimal path in the area of ​​interest to the user of the dual-axle steering unmanned mining truck, and plans the vehicle driving state based on the optimal path using the simulated annealing algorithm; The LiDAR mapping module generates laser odometry information based on the forward LiDAR, providing the vehicle's relative position and posture information. It generates a map based on point cloud feature information matching based on the backward LiDAR, performs loop correction optimization on the laser odometry information of the forward LiDAR, and performs point cloud degrounding, screening, clustering, and matching on the generated map to ultimately form a three-dimensional map. The measurement and calculation results of the two LiDARs are uploaded to the cloud, which matches the measurement and calculation results of the two based on the time frame to form a complete map. The LiDAR positioning module uses the LiDAR point cloud feature information to match the normally distributed point cloud with the complete map on the cloud, generates probability information, and determines the positioning position of the LiDAR. Specifically, the complete point cloud map on the cloud is obtained as a priori map, and the forward LiDAR A performs point cloud normally distributed feature matching according to the priori map to obtain the current coarse positioning A of the vehicle body. Based on the information of the coarse positioning A, the backward LiDAR B performs point cloud feature matching to output the fine positioning B.

2. A real-time mapping method using the laser radar real-time mapping and positioning system according to claim 1, characterized in that: The following steps are involved: S1 users calibrate the parameters of the front and rear LiDARs of the dual-axle steering unmanned mining truck based on their internal and external parameters and positional relationships. S2 The user selects an area of ​​interest from the map and selects a path based on the area of ​​interest as the planned path; S3 activates the dual-axle steering unmanned mining truck. After reaching the designated starting point of the planned path, the front and rear lidars are turned on. The S4 dual-axle steering unmanned mining truck travels along the planned route. During driving, the forward LiDAR A generates laser odometry information, while the rear LiDAR B collects map point cloud information. The laser odometry information from the forward LiDAR A is then loop-corrected. A map is then constructed based on the point cloud features in the point cloud information. The measurement and calculation results of the two LiDARs are fused to generate a complete point cloud map, which is then uploaded to the cloud. In step S1, the user inputs the measured extrinsic parameter values, and the computer performs iterative optimization of normal distribution feature matching based on the extrinsic parameter values ​​and the point clouds generated by the two laser radars in a stationary state, so that the point clouds of the two laser radars can eventually reach a completely matched state.

3. The real-time mapping method according to claim 2, characterized in that: The planned path in step S2 is obtained by using the shortest path algorithm.

4. The real-time mapping method according to claim 2, characterized in that: The laser odometry information in step S4 is generated by the point cloud curvature feature information collected by the forward laser radar.

5. A positioning method based on the real-time mapping method according to any one of claims 2 to 4, characterized in that: The following steps are also included: S5: After the dual-axle steering unmanned mining truck reaches the end of the planned path, it switches the unmanned mining truck to the forward and backward directions and returns along the original planned path; S6's forward-facing LiDAR A and rear-facing LiDAR B retrieve the cloud point cloud map, match the point cloud according to the normal distribution change, and generate solution values. The solution values ​​are transformed relative to the center point to generate positioning A and positioning B, respectively. The final vehicle positioning is generated based on weighted calculation.

Images