Unmanned mine car obstacle avoidance method, system and equipment and storage medium

Abstract

The invention relates to the technical field of unmanned driving, in particular to an unmanned mine car obstacle avoidance method, system and device and a storage medium, and the method comprises the steps: collecting road condition information, building a map, and planning a global path of a mine car; collecting real-time data, and performing weighted fusion according to the confidence coefficient weight to obtain a fusion result; updating the grid state of the map according to the fusion result; the real-time pose of the mine car is calculated according to fusion of the laser radar and the chassis encoder; and adjusting the local path of the mine car according to the grid state and the real-time pose of the mine car, and implementing an obstacle avoidance strategy. The obstacle avoidance efficiency of the unmanned mine car is improved, and potential safety hazards are reduced.

Description

Technical Field

[0001] This disclosure relates to the field of unmanned driving technology, and in particular to an obstacle avoidance method, system, device and storage medium for unmanned mining trucks. Background Technology

[0002] Unmanned mining trucks are intelligent systems that integrate sensing technology, vehicle-side control algorithms, and cloud-based collaboration to achieve fully automated operations throughout the entire mining process.

[0003] Current unmanned mining truck technology relies on a single lidar or vision sensor. Dust and low light cause distortion of perception data, making it impossible to reliably identify dynamic targets such as falling rocks, temporary equipment, and personnel. The use of pre-built fixed mine road maps cannot update sudden obstacles in the mine road in real time, resulting in delayed path planning. Therefore, the obstacle avoidance efficiency of existing unmanned mining truck technology is low and there are safety hazards. Summary of the Invention

[0004] To address the aforementioned technical problems, this disclosure provides a method, system, device, and storage medium for obstacle avoidance by unmanned mining vehicles, wherein the method includes: Collect road condition information to build a map and plan the global path for the mining trucks; Real-time data is collected and weighted according to confidence levels to obtain the fusion result; The raster state of the map is updated based on the fusion result; The real-time pose of the mining truck is calculated based on the fusion of lidar and chassis encoder data. The mine car's local path is adjusted based on the grid state and the real-time pose of the mine car to implement an obstacle avoidance strategy.

[0005] Further, updating the raster state of the map based on the fusion result includes: If the fusion result detects that the obstacle has existed for a time greater than or equal to a preset time, then the grid state of the map is updated; If the fusion result detects that the obstacle has existed for less than a preset time, the current grid state of the map is maintained. If the fusion result detects an obstacle size greater than or equal to a preset avoidance space, the grid state of the map is updated; If the fusion result detects an obstacle smaller than a preset avoidance space, the grid state of the map is maintained.

[0006] Further, after the step of calculating the real-time pose of the mining truck based on the fusion of lidar and chassis encoder, the following steps are included: The positioning covariance is calculated based on the real-time pose of the mining vehicle. The mine car is adjusted to the positioning marker based on the positioning covariance. The coordinates of the positioning marker are obtained and compared with the positioning result of the lidar to obtain the pose deviation value of the mine car; The current pose of the mine car is corrected based on the pose deviation value, and the covariance is reset.

[0007] Further, adjusting the mine car to the positioning marker based on the positioning covariance includes: If the covariance is less than a preset threshold, the current real-time pose of the mining truck is maintained. If the covariance is greater than or equal to a preset threshold, the mine car is corrected to travel to the positioning marker.

[0008] Furthermore, the implementation of the obstacle avoidance strategy includes: Based on the real-time data, the target type is obtained; Calculate the safe distance to the target based on the real-time data and the parameters of the mining truck; Braking or adjusting the speed of the mine car or replanning the local path may be performed based on the target type and the safety distance.

[0009] Furthermore, it also includes: The movement status of the mining truck is monitored based on its global path, and feedback is provided. The feedback is analyzed and optimized, and the mining truck data is cleared.

[0010] This disclosure also provides an obstacle avoidance system for unmanned mining trucks, including: The data acquisition module is used to collect road condition information to build a map and plan the global path of the mining truck. The fusion module is used to collect real-time data and perform weighted fusion based on confidence level to obtain the fusion result; An update module is used to update the raster state of the map based on the fusion result; The calculation module is used to calculate the real-time pose of the mining truck based on the fusion of the lidar and the chassis encoder. The obstacle avoidance module is used to adjust the local path of the mine car according to the grid state and the real-time pose of the mine car, and implement the obstacle avoidance strategy.

[0011] Furthermore, it also includes: The pose module is used to calculate the positioning covariance based on the real-time pose of the mining truck. The adjustment module is used to adjust the mine car to the positioning marker according to the positioning covariance; The comparison module is used to obtain the coordinates of the positioning marker, compare them with the positioning result of the lidar, and obtain the pose deviation value of the mine car. The reset module is used to correct the current pose of the mine car based on the pose deviation value and reset the covariance.

[0012] This disclosure also provides a computer device, including a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the transmission tower fault prediction method.

[0013] This disclosure also provides a computer-readable storage medium storing computer-readable instructions, which, when executed by a processor, implement the steps of the transmission tower fault prediction method.

[0014] The technical solution provided in this disclosure has the following advantages compared with the prior art: This invention improves global path planning by real-time road condition information collection and map construction, making it more consistent with the structural characteristics of mine tunnels. It significantly enhances the reliability and anti-interference capability of perception results through multi-source data fusion based on confidence weights. Regularized dynamic updates of the grid state ensure both map stability and real-time performance. Fusion positioning of LiDAR and chassis encoders, along with a positioning covariance correction mechanism, ensures the mine truck's pose remains consistently accurate and traceable. By combining grid state and real-time pose with local path adjustment and obstacle avoidance strategies, the mine truck can maintain stable and safe autonomous driving in complex, dynamic, and even harsh underground mining environments. This invention achieves data interoperability and collaborative optimization among the mine truck path planning, environmental perception, positioning, and obstacle avoidance modules, reducing the risk of misjudgment, minimizing replanning, and improving driving continuity, obstacle avoidance efficiency, and overall transportation operation safety. Attached Figure Description

[0015] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0016] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the obstacle avoidance method for unmanned mining trucks according to an embodiment of this disclosure; Figure 2 As described in the embodiments of this disclosure Figure 1 A schematic diagram of the method for updating the raster state in the middle; Figure 3 This is a schematic diagram of the obstacle avoidance system for unmanned mining trucks as described in an embodiment of this disclosure. Detailed Implementation

[0018] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0019] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0020] Figure 1 This is a schematic diagram of the obstacle avoidance method for unmanned mining trucks described in an embodiment of this disclosure; as shown... Figure 1 As shown, an obstacle avoidance method for unmanned mining trucks includes: Step S1: Collect road condition information to build a map and plan the global path for the mining trucks; In this step, during system initialization, the static map base unit in the mine tunnel dynamic map construction and update module is invoked to load the fixed structures of the mine tunnel (such as support positions, track directions, and intersection coordinates) from the pre-stored mine tunnel design drawings, establishing a static layer in a hybrid form of grid and feature points. Simultaneously, the sensor units (LiDAR, millimeter-wave radar, infrared thermal imager, dust concentration sensor, and road surface humidity sensor) of the underground multi-source environmental perception fusion module are activated to collect initial road condition data. After the data is synchronized, denoised (point cloud statistical filtering, Kalman filtering, etc.) and coordinate transformed by the data preprocessing unit, the initial state of the dynamic layer is updated to reflect the current distribution of obstacles. Subsequently, the global path planning unit in the hierarchical path planning module uses the improved A* algorithm (A* Search Algorithm) on the static layer and calculates the global path from the starting point to the ending point based on the global path cost function. The generated global path includes key nodes (intersections, positioning markers, etc.) and is written to the path database for use in local planning. By solidifying the design drawings into a static layer and integrating them with the initial road conditions collected in real time, the feasibility of the global path is ensured under the consideration of fixed constraints of the mine road and initial dynamic obstacles. By introducing the costs of load and dynamic obstacles, the path selection is coupled with the physical constraints (load and turning ability) of the mine car, thereby improving operational safety and task completion efficiency and reducing the frequency of replanning afterward.

[0021] Step S2: Collect real-time data and perform weighted fusion based on confidence level to obtain the fusion result; In this step, during system operation, the sensor units continuously collect raw data: the lidar outputs a 3D point cloud of the foreground, the millimeter-wave radar outputs target distance and velocity, the infrared thermal imager outputs a thermal map for personnel detection, and the dust / humidity sensor outputs environmental interference parameters. The data preprocessing unit performs time synchronization, coordinate transformation, and denoising on each type of raw data (statistical filtering of the lidar point cloud removes low-density points, Kalman filtering is used to smooth the millimeter-wave velocity, and background modeling and threshold segmentation are used to extract contours from the infrared image). After preprocessing, the measurement values ​​of each sensor are obtained. The multi-source fusion unit uses a confidence-weighted fusion algorithm to weight the measurement values ​​of each sensor, where the weights are dynamically adjusted according to environmental parameters (dust concentration, ambient light, road surface humidity, etc.) (e.g., reducing the lidar weight and increasing the millimeter-wave radar weight when the dust concentration is high; increasing the infrared thermal image weight when the light is low). The fusion result includes obstacle category (personnel / mobile equipment / static obstacle), position, distance, velocity, and confidence index. To prevent misjudgment due to transient noise, the fusion unit can use historical window filtering for the same target and output statistics on the target's existence time and size. By dynamically adjusting the confidence level of the coupling between environmental perception and environmental parameters, the reliability of perception under complex conditions such as dust and low light in mines is significantly improved. The confidence level of the fusion result makes it easier for downstream modules (localization, map update, obstacle avoidance decision-making) to make conservative or aggressive decisions based on the confidence level, thereby reducing false triggering and missed detection rates.

[0022] Step S3: Update the raster state of the map based on the fusion result; In this step, the dynamic update unit receives the fusion results and maps the detected obstacle information onto the dynamic layer grid of the map. The update follows a triggering logic: when a grid is continuously observed to be in an "occupied" state for a number of times greater than or equal to a preset threshold, or when the occupation duration is greater than or equal to a preset time and the obstacle size is greater than or equal to the minimum avoidance space for the mining truck, the grid is marked as "occupied"; otherwise, it is marked as "idle". Simultaneously, the timestamp of grid occupation, obstacle type, and confidence level are recorded for historical backtracking. The dynamic and static layers adopt a hierarchical storage structure. The dynamic layer is a real-time cache, and after updating, it is synchronously pushed to the positioning (for matching) and path planning modules (for local planning) through the map storage and retrieval unit. To reduce update overhead and false alarm rate, an occupation probability or Bayesian update strategy can be introduced into the dynamic layer, and priority can be set for critical grids (grids closer to the vehicle path are updated first). Employing a grid update logic with persistence and size thresholds can filter out transient noise and reduce the impact of false alarms on planning; hierarchical storage makes static geometric information stable and efficient, while the dynamic layer can reflect temporary obstacles in a timely manner, thereby ensuring that path planning and localization have both stable priors and real-time response capabilities, improving the accuracy of obstacle avoidance decisions and the robustness of the system.

[0023] Step S4: Calculate the real-time pose of the mining truck based on the fusion of the lidar and the chassis encoder; In this step, the core localization unit uses a laser SLAM (Simultaneous Localization and Mapping) algorithm (utilizing static features of the mine roadway scanned by lidar, such as support spacing and intersection shape) to extract and match features from the point cloud, generating a real-time pose solution. Simultaneously, it collects data from the chassis encoder odometer to compensate for short-term occlusion or partial radar obstruction. The laser SLAM results are fused with encoder data using an extended Kalman filter (EKF) or graph-based SLAM to output a smooth pose estimate, and the localization covariance matrix is ​​calculated. The localization reliability assessment unit compares the individual variance with a preset threshold: when the covariance is less than the threshold, SLAM and encoder pose are continued; when the covariance is greater than or equal to the threshold, the pose correction unit is triggered: the mine truck is controlled to approach the nearest positioning marker (e.g., a reflector), and the absolute coordinates of the marker are read via visual recognition or a card reader, compared with the current SLAM pose, and the pose is corrected using the difference while resetting the covariance. All pose information and confidence scores are simultaneously transmitted back to the path planning and obstacle avoidance modules. The fusion of laser SLAM and encoders not only utilizes environmental features to provide high-precision positioning, but also uses odometer constraints to improve short-term robustness. The covariance-driven correction strategy can automatically restore the absolute reference when the positioning quality deteriorates, avoiding the accumulation of positioning errors caused by long-term drift, thereby ensuring the reliability and safety of path tracking and obstacle avoidance judgment.

[0024] Step S5: Adjust the local path of the minecart based on the grid status and the real-time pose of the minecart, and implement obstacle avoidance strategy.

[0025] In this step, the local path planning unit uses the current key nodes of the global path as a benchmark and takes the look-ahead distance (e.g., 50–100m) as the local planning range. It calls the dynamic layer grid information and the current pose to perform improved RRT (Rapidly-exploring Random Tree) planning, generating a locally feasible path that meets the kinematic constraints of the mining truck (minimum turning radius, maximum climbing angle). The path is then smoothed by the path optimization unit using B-spline (BasisSpline). Based on this, the multi-scenario obstacle avoidance decision unit formulates differentiated strategies based on obstacle classification and confidence, grid occupancy, and the minimum safe distance calculated by the safe distance calculation unit, taking into account the road friction coefficient and gravity: for personnel targets, it performs deceleration and detour or emergency braking according to the distance threshold; for mobile devices, it negotiates passing / following strategies through vehicle-to-vehicle communication and plans passing points; for static obstacles, it determines whether to detour or trigger global replanning based on the size and the mining truck's avoidance space. Once a decision is generated, the path correction unit sends the corrected local path and velocity profile to the mine car motion state monitoring and control module. Adaptive PID control (Proportional-Integral-Derivative Controller) (PID parameters dynamically adjusted according to load) generates chassis control commands and sends them to the execution unit. The motion control module provides real-time feedback on the actual pose and execution status to the local planning module. If the deviation exceeds a threshold, it triggers replanning or a safe stop. The grid-based and pose-based closed-loop local planning and differentiated obstacle avoidance strategy can quickly respond to dynamic obstacles while maintaining consistency with the global path, ensuring both driving efficiency and reducing collision risk. Coupling safety distance with vehicle dynamics (load, road friction) makes obstacle avoidance decisions more consistent with actual braking capabilities, improving the reliability of safety boundaries in real-world scenarios. Vehicle-to-vehicle communication and cooperation strategies further enhance traffic efficiency and collaborative performance in multi-vehicle scenarios.

[0026] Figure 2 As described in the embodiments of this disclosure Figure 1 A schematic diagram of the method for updating the raster state; as shown. Figure 2 As shown, in another embodiment of this disclosure, step S3, updating the grid state of the map according to the fusion result, includes: step S31, if the fusion result detects that the obstacle exists for a time greater than or equal to a preset time, then update the grid state of the map; step S32, if the fusion result detects that the obstacle exists for a time less than the preset time, then maintain the current grid state of the map; step S33, if the fusion result detects that the obstacle size is greater than or equal to a preset avoidance space, then update the grid state of the map; step S34, if the fusion result detects that the obstacle size is less than the preset avoidance space, then maintain the grid state of the map.

[0027] Specifically, the system obtains the fusion results from the multi-source sensor fusion module and extracts parameters such as obstacle existence time and obstacle size. The obstacle existence time is the cumulative time during which the fused target is stably identified within a continuous detection cycle. When the fusion result detects an obstacle existence time greater than or equal to a preset time, the system maps the obstacle to a dynamic map grid and updates the grid status to "occupied." When the fusion result detects an obstacle existence time less than the preset time, the system determines that the current detection may be noise or transient interference, thus maintaining the current map grid status to avoid falsely triggering occupancy markers. Furthermore, the system determines the obstacle size in the fusion result: if the obstacle size is greater than or equal to a preset avoidance space, the corresponding grid is updated to "occupied" to prompt the subsequent path planning module to execute a detour strategy; if the obstacle size is less than the preset avoidance space, the system maintains the current grid status to reduce invalid disturbances to the planned path. By using both obstacle existence time and obstacle size as the basis for grid updates, the interference of instantaneous false detections and small, insignificant objects on map updates can be effectively suppressed, thereby improving the stability and reliability of dynamic maps and enabling the planning module to perform safe obstacle avoidance based on a more accurate obstacle distribution.

[0028] In another embodiment of this disclosure, step S4, after the step of calculating the real-time pose of the mining car based on the fusion of the lidar and the chassis encoder, includes: calculating the positioning covariance based on the real-time pose of the mining car; adjusting the mining car to the positioning marker based on the positioning covariance; obtaining the coordinates of the positioning marker, comparing them with the lidar positioning result, and obtaining the pose deviation value of the mining car; correcting the current pose of the mining car based on the pose deviation value of the mining car, and resetting the covariance.

[0029] Specifically, the system first calculates the positioning covariance based on the real-time pose of the mining truck to assess the uncertainty of the current pose estimation. The positioning covariance reflects the impact of lidar matching quality, odometer cumulative error, and environmental observability on pose accuracy. When the positioning covariance reaches a level that does not meet the positioning accuracy requirements, the system triggers a positioning correction operation: the mining truck is adjusted to a preset positioning marker position based on the positioning covariance, the coordinates of the positioning marker are obtained through a visual recognition device or reflector, and the coordinates are compared with the lidar positioning results to obtain the mining truck pose deviation value. The system then corrects the current mining truck pose based on this pose deviation value, while resetting the positioning covariance to its initial value to restore the global accuracy of the positioning system. By performing absolute calibration based on the positioning marker when the positioning covariance is abnormal, the cumulative error in the lidar and encoder fusion process can be effectively suppressed, avoiding obstacle avoidance failure or path deviation caused by long-term drift of the mining truck pose, thereby significantly improving the stability and safety of the overall positioning system.

[0030] In another embodiment of this disclosure, adjusting the mine car to the positioning marker based on the positioning covariance includes: if the covariance is less than a preset threshold, maintaining the current real-time pose of the mine car; if the covariance is greater than or equal to the preset threshold, correcting the mine car to travel to the positioning marker.

[0031] Specifically, the system monitors the real-time positioning covariance and compares it with a preset threshold. When the covariance is less than the preset threshold, it indicates that the reliability of the current real-time pose of the mining truck is high, so the system maintains the current real-time pose of the mining truck and does not perform any additional adjustments. When the covariance is greater than or equal to the preset threshold, the system determines that the current positioning quality has deteriorated to the point where calibration is required. At this time, a correction strategy is triggered, controlling the mining truck to travel along a safe path to a positioning marker with known coordinates, so that absolute position correction can be completed through the positioning marker. The covariance threshold determination mechanism enables real-time monitoring of positioning accuracy and triggers automatic correction when positioning reliability is insufficient, effectively avoiding positioning drift caused by environmental complexity, dust obstruction, feature sparsity, and other factors, and improving the adaptability and robustness of the positioning system.

[0032] In another embodiment of this disclosure, step S5 implements an obstacle avoidance strategy, including: obtaining the target type based on real-time data; calculating the safe distance from the target based on real-time data and mine car parameters; and braking or adjusting the mine car speed or replanning the local path based on the target type and the safe distance.

[0033] Specifically, the system identifies target types (personnel, mobile devices, static obstacles, etc.) based on real-time data and calculates the safe distance to the target by combining fusion results, radar ranging results, and the current motion parameters of the mining truck. The safe distance can be calculated comprehensively based on the mining truck's current speed, braking capacity (maximum deceleration), tire-ground friction coefficient, and system response time. The system then formulates obstacle avoidance behavior based on the target type and safe distance: if the target is a person and the distance is less than the corresponding safe distance, a braking strategy is prioritized; if the target is a mobile device and the distance is within the danger zone, the mining truck speed is adjusted or a passing / yielding strategy is implemented; if the target is a static obstacle, the system determines whether to trigger a replanning of the local path based on the safe distance. After this strategy is executed, the system sends a new speed command or local path to the motion control module to ensure the mining truck safely avoids obstacles. By coupling target type identification, safe distance calculation, and obstacle avoidance decision-making, differentiated obstacle avoidance behaviors can be achieved for different scenarios, making obstacle avoidance decisions more aligned with the mining truck's dynamic characteristics and scenario safety requirements, thereby improving the safety performance and scheduling efficiency of the mining truck under complex working conditions.

[0034] In another embodiment of this disclosure, the method further includes: monitoring the movement status of the mining truck based on the global path of the mining truck and providing feedback; analyzing and optimizing based on the feedback, and clearing the mining truck data.

[0035] Specifically, the system continuously monitors the operation of mining trucks along the global path, including position deviation, speed status, path tracking error, and obstacle avoidance execution. Monitoring data is transmitted back to the analysis module in real time. Based on the feedback, the analysis module determines whether the mining trucks have issues such as path deviation, abnormal speed, accumulated positioning errors, or insufficient obstacle avoidance strategy execution. Based on the analysis results, it optimizes local planning, positioning weights, or control parameters to improve overall system performance. When a global task is completed or path planning is regenerated, the system clears the mining truck's historical data, including historical positioning data, local path cache, and historical fusion information, to avoid historical data accumulating errors or interfering with the next task. Continuous monitoring and feedback analysis based on the global path forms a closed-loop control mechanism, improving the responsiveness and anomaly handling capabilities of the mining truck scheduling system. The data clearing mechanism after a task is completed effectively reduces the risk of data interference, maintaining the long-term stability and reliability of the system.

[0036] In another embodiment of this disclosure, a control method for an unmanned mining truck path planning and obstacle avoidance control system includes: System initialization and sensor calibration: Start all system modules and calibrate the sensors. LiDAR: Uses a support structure of known dimensions within the mine tunnel to calibrate distance measurement deviations; Millimeter-wave radar: calibrates speed measurement accuracy using a test vehicle traveling at a fixed speed within the mine tunnel; Infrared thermal imager: calibrates personnel identification thresholds using a standard heat source within the mine tunnel; Load the pre-stored static map of the mine tunnel, including the location of the support and the coordinates of the intersection, and initialize the starting pose of the mine car by obtaining the initial positioning marker; Set task parameters: start point, end point, load capacity threshold, and maximum driving speed.

[0037] Multi-source environmental sensing and data fusion: real-time acquisition of mine tunnel environmental data by various sensors. LiDAR: Scans the three-dimensional point cloud of obstacles within a 100m range in front of the mine tunnel; Millimeter-wave radar: detects the speed and distance of dynamic targets within a 200m range in front; Infrared thermal imager: Identifies personnel targets within a 50m range in front; Dust concentration or road surface humidity sensor: collects environmental interference parameters; The data preprocessing unit processes the raw data as follows: LiDAR point cloud: Use statistical filtering to remove dust and noise points, and retain areas where the point cloud density is greater than or equal to the threshold. Millimeter-wave radar data: velocity fluctuations are smoothed using Kalman filtering, with the filtering period matching the sensor sampling period; Multi-source fusion unit according to formula The calculation and fusion results are output, including obstacle type, distance, and speed information.

[0038]

[0039] in, The fused environmental perception results; Sensor number, =1 indicates a lidar system. =2 indicates millimeter-wave radar. =3 is an infrared thermal imager; : No. The confidence weight of each sensor, with a value range of 0-1, is dynamically adjusted by environmental parameters.

[0040] When dust concentration is high: reduce, Increase; Under low light conditions: rise, reduce; : No. The original measurement values ​​of each sensor, and the obstacle distance after preprocessing and noise reduction.

[0041] The dynamic map construction and updating of the mine tunnel involves the static map basic unit calling pre-stored static structure data of the mine tunnel to generate the initial map; The dynamic update unit receives the fusion results and determines whether to trigger a map update: If a new obstacle is detected, and its existence time is greater than or equal to the system response time, and its size is greater than or equal to the minimum clearance space for the minecart, the corresponding grid will be marked as "occupied". If the original obstacle disappears, the corresponding grid will be marked as "empty"; The map storage unit updates the dynamic layer data and pushes it synchronously to the positioning and route planning modules.

[0042] Real-time positioning and pose correction of the mining truck; the core positioning unit runs a laser SLAM algorithm. LiDAR scans the static features of the mine tunnel and matches them with the static layer of the dynamic map; By combining chassis encoder data, the mileage of the mining truck is recorded, and the real-time pose of the mining truck is calculated. , and ); The positioning reliability assessment unit calculates the SLAM positioning covariance: If the covariance is less than the threshold: continue to use the SLAM localization results; If the covariance is greater than or equal to the threshold: trigger emergency correction and control the mine car to travel to the nearest positioning marker; Position correction unit identifies positioning markers: Obtain the absolute coordinates of the marker, compare them with the SLAM localization results, and calculate the pose deviation. Correct the current pose of the minecart using the deviation value and reset the SLAM covariance.

[0043] Hierarchical path planning execution, with the global path planning unit based on the static layer of the dynamic map, according to the formula Calculate path cost: in, : The total cost of the global path, used to rank paths by quality; the smaller the value, the better the path. , , , The weighting coefficients of each cost item. Adjustments will be made based on the requirements of the mining truck mission. Distance cost, the ratio of path length to shortest path length. ≥1; Turning cost: The ratio of the sum of all turning angles along the path to the sum of the maximum allowed turning angles for the minecart. ≥0; Load cost: The ratio of the actual load of the mining truck to its maximum load, 0 ≤ ≤1, the greater the load capacity The larger; Dynamic obstacle cost: The ratio of the number of dynamic obstacles around the path to the total obstacle capacity of the mine tunnel, 0 ≤ ≤1; Input the starting and ending coordinates, traverse the mine tunnel grid using the improved A algorithm, and select the total cost. The shortest path is used as the global path; The global path contains multiple key nodes (such as intersections and location markers), which are stored in the path planning unit; Local path planning units are based on dynamic layers of a dynamic map: Starting from the current node of the global path and ending 50-100m ahead, a local path is generated using an improved RRT algorithm. Local paths must satisfy the kinematic constraints of the mine car, with the minimum turning radius ≥ the mine car's design value and the maximum climbing angle ≤ the mine car's maximum climbing capacity; The path optimization unit smooths local paths: B-spline curves are used to fit path points to eliminate abrupt curvature changes, and the rate of curvature change is less than or equal to the allowable value for the mine car. The path speed limit is adjusted according to the load capacity of the mining car; when fully loaded, the speed on curves is reduced by 20%-30%.

[0044] Multi-scenario obstacle avoidance decision-making and path correction, obstacle classification unit classifies the fusion results: Personnel target: Identification of human body contour features using infrared thermal imagers; Mobile devices: Identified using the speed characteristics of millimeter-wave radar and the shape characteristics of lidar; Static obstacles: identified by the fixed positional features of LiDAR; The safe distance calculation unit is based on the formula. calculate : Enter the current speed of the minecart Road surface moisture (calculated) ), load capacity (affected) ), output ; in, The minimum safe distance between a mining car and an obstacle; : Current speed of the mining truck; System response time, including sensing, decision-making, and control delays; The maximum deceleration of a mining car is determined by the road surface moisture and the load weight. ,in The coefficient of friction for wet and slippery road surfaces. reduce; It is the acceleration due to gravity; : Safety redundancy distance to avoid collision risks caused by measurement errors; Obstacle avoidance decision-making unit formulates strategy: Personnel target: If the distance is ≥2 Control the mine car to decelerate (acceleration ≤ And detour; if the distance is <2 This triggered emergency braking; Mobile devices: Obtain the other party's path through vehicle-to-vehicle communication; if there is a need for passing, plan the passing point and adjust their own speed to the passing speed. Static obstacles: If the size is less than or equal to the clearance space for the minecart, a local detour path is regenerated; if the size is greater than the clearance space, a global path replanning is triggered. The path correction unit updates the local path and pushes the corrected path to the motion control module.

[0045] Mine car motion control and status feedback: The status monitoring unit collects mine car parameters in real time. Driving speed and acceleration are calculated by fusing the encoder with the IMU (Inertial Measurement Unit). Braking pressure (collected by a pressure sensor), load capacity (collected by a weighing sensor); If the parameters are abnormal (such as braking pressure < safety threshold), an early warning will be triggered immediately, and path tracking will be suspended. The velocity planning unit generates a trapezoidal velocity curve: Based on the optimized path, set an acceleration segment (acceleration ≤ 80% of the total speed), constant speed section (speed ≤ path speed limit), deceleration section (acceleration ≥ - ); The control algorithm unit executes adaptive PID control: Calculate the deviation between the actual position and orientation of the mining truck and the planned path, namely the lateral deviation and the heading angle deviation; Adjust the PID parameters according to the load capacity, increasing them by 10%-20% when fully loaded to improve control response; The deviation is translated into control commands, including throttle opening, brake pedal travel, and steering angle; The execution unit sends control commands to the chassis ECU (Electronic Control Unit) to drive the mining truck; at the same time, it feeds back the actual driving status to the path planning and obstacle avoidance modules to achieve closed-loop control.

[0046] Emergency monitoring and fault handling: The emergency safety assurance module monitors system faults in real time. Sensor failure: If the lidar has no data, switch to a combination of millimeter-wave radar and infrared thermal imager for sensing. Communication interruption: Activate the local emergency map and control the minecart to travel at low speed to the nearest escape lane; Collision risk: If the obstacle avoidance decision is delayed, resulting in a distance < This triggers emergency braking; when the braking pressure is ≥90% of the maximum pressure, the warning device is activated. The fault reporting unit reports the fault type and mine car location to the dispatch center through the mine tunnel wired network; After troubleshooting, the system receives a recovery command from the dispatch center and continues to complete the task.

[0047] Task completion and system reset: When the mine car reaches the destination (position deviation ≤ threshold), a task completion signal is output. Reset each module: Clear dynamic layer data from dynamic map; The positioning module resets the pose record; The motion control module stops outputting control commands; Store the data for this task, including path records, obstacle avoidance attempts, and fault information, for system optimization.

[0048] Figure 3 This is a schematic diagram of the obstacle avoidance system for unmanned mining trucks described in an embodiment of this disclosure; as follows: Figure 3 As shown, this disclosure also provides an obstacle avoidance system for unmanned mining trucks, including: The data acquisition module 401 is used to collect road condition information, build a map, and plan the global path for mining trucks; The fusion module 402 is used to collect real-time data and perform weighted fusion according to the confidence level to obtain the fusion result; Update module 403 is used to update the raster state of the map based on the fusion result; Calculation module 404 is used to calculate the real-time pose of the mining truck based on the fusion of lidar and chassis encoder. The obstacle avoidance module 405 is used to adjust the local path of the mine car according to the grid status and the real-time pose of the mine car, and implement the obstacle avoidance strategy.

[0049] In another embodiment of this disclosure, it further includes: The pose module is used to calculate the positioning covariance based on the real-time pose of the mining truck. The adjustment module is used to adjust the mine car to the positioning marker based on the positioning covariance; The comparison module is used to obtain the coordinates of the positioning marker, compare them with the positioning results of the lidar, and obtain the pose deviation value of the mine car. The reset module is used to correct the current pose of the mine car based on the pose deviation value and reset the covariance.

[0050] In another embodiment of this disclosure, This disclosure also provides a computer device, including a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the transmission tower fault prediction method.

[0051] This disclosure also provides a computer-readable storage medium storing computer-readable instructions, which, when executed by a processor, implement the steps of a transmission tower fault prediction method.

[0052] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0053] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for obstacle avoidance by unmanned mining vehicles, characterized in that, Collect road condition information to build a map and plan the global path for the mining trucks; Real-time data is collected and weighted according to confidence levels to obtain the fusion result; The raster state of the map is updated based on the fusion result; The real-time pose of the mining truck is calculated based on the fusion of lidar and chassis encoder data. The mine car's local path is adjusted based on the grid state and the real-time pose of the mine car to implement an obstacle avoidance strategy.

2. The obstacle avoidance method for unmanned mining vehicles according to claim 1, characterized in that, Updating the raster state of the map based on the fusion result includes: If the fusion result detects that the obstacle has existed for a time greater than or equal to a preset time, then the grid state of the map is updated; If the fusion result detects that the obstacle has existed for less than a preset time, the current grid state of the map is maintained. If the fusion result detects an obstacle size greater than or equal to a preset avoidance space, the grid state of the map is updated; If the fusion result detects an obstacle smaller than a preset avoidance space, the grid state of the map is maintained.

3. The obstacle avoidance method for unmanned mining vehicles according to claim 1, characterized in that, After the step of calculating the real-time pose of the mining truck based on the fusion of lidar and chassis encoder, the following is included: The positioning covariance is calculated based on the real-time pose of the mining vehicle. The mine car is adjusted to the positioning marker based on the positioning covariance. The coordinates of the positioning marker are obtained and compared with the positioning result of the lidar to obtain the pose deviation value of the mine car; The current pose of the mine car is corrected based on the pose deviation value, and the covariance is reset.

4. The obstacle avoidance method for unmanned mining vehicles according to claim 3, characterized in that, The step of adjusting the mine car to the positioning marker according to the positioning covariance includes: If the covariance is less than a preset threshold, the current real-time pose of the mining truck is maintained. If the covariance is greater than or equal to a preset threshold, the mine car is corrected to travel to the positioning marker.

5. The obstacle avoidance method for unmanned mining vehicles according to claim 1, characterized in that, The implementation of the obstacle avoidance strategy includes: Based on the real-time data, the target type is obtained; Calculate the safe distance to the target based on the real-time data and the parameters of the mining truck; Braking or adjusting the speed of the mine car or replanning the local path may be performed based on the target type and the safety distance.

6. The obstacle avoidance method for unmanned mining vehicles according to claim 1, characterized in that, Also includes: The movement status of the mining truck is monitored based on its global path, and feedback is provided. The feedback is analyzed and optimized, and the mining truck data is cleared.

7. An unmanned mining truck obstacle avoidance system, characterized in that, include: The data acquisition module is used to collect road condition information to build a map and plan the global path of the mining truck. The fusion module is used to collect real-time data and perform weighted fusion based on confidence level to obtain the fusion result; An update module is used to update the raster state of the map based on the fusion result; The calculation module is used to calculate the real-time pose of the mining truck based on the fusion of the lidar and the chassis encoder. The obstacle avoidance module is used to adjust the local path of the mine car according to the grid state and the real-time pose of the mine car, and implement the obstacle avoidance strategy.

8. The unmanned mining truck obstacle avoidance system according to claim 7, characterized in that, Also includes: The pose module is used to calculate the positioning covariance based on the real-time pose of the mining truck. The adjustment module is used to adjust the mine car to the positioning marker according to the positioning covariance; The comparison module is used to obtain the coordinates of the positioning marker, compare them with the positioning result of the lidar, and obtain the pose deviation value of the mine car. The reset module is used to correct the current pose of the mine car based on the pose deviation value and reset the covariance.

9. A computer device, characterized in that, The method includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the transmission tower fault prediction method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the transmission tower fault prediction method as described in any one of claims 1 to 6.

Images