Autonomous obstacle avoidance and turning track planning method for monorail crane equipment

Abstract

The invention discloses a turning track planning method for monorail crane equipment, and particularly relates to the field of transportation equipment path planning, the method comprises the following steps: a sensor system mounted on a monorail crane detects distance data of the surrounding environment of the monorail crane equipment in real time; based on the distance data of the surrounding environment of the monorail crane equipment, generating a two-dimensional model of the surrounding environment of the monorail crane equipment through a data fusion technology; and according to the two-dimensional model of the surrounding environment of the monorail crane equipment, the turning path of the monorail crane is calculated in advance. According to the method, the monorail crane equipment autonomously, safely and stably completes optimal path turning in a dynamic complex environment, and the operation efficiency and the environment adaptability of the monorail crane equipment are improved.

Description

Technical Field

[0001] This invention relates to the field of transportation equipment path planning technology, and more specifically, to a method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes. Background Technology

[0002] Monorail cranes are widely used in industrial production, logistics transportation, and the transportation of equipment and materials in coal mines due to their advantages such as high reliability, strong transportation capacity, low cost, convenient installation, and small turning radius. They can achieve direct transportation and reduce the number of transfers.

[0003] Existing publicly available literature 1 (Design and Analysis of Speed-Adaptive Mountain Monorail Transporter, 2022) designs a purely mechanical, speed-adaptive monorail transporter, such as... Figure 2 As shown, the monorail transport machine consists of a four-wheeled vehicle body, a guiding mechanism, and a speed adaptive adjustment system. The four-wheeled vehicle body and guiding mechanism include wheels, guide wheels, guide rods, and a width adjustment mechanism, which can adapt to different track widths and maintain stable contact with the inner wall of the track. The speed adaptive adjustment system includes a speed transmission device and a braking device. However, the monorail transport machine is prone to collisions with obstacles in the complex working environment of mountainous terrain.

[0004] Existing publicly available literature 2 (Exploring the key design points of current coal mine monorail transportation systems, 2024) proposes a monorail operating roadway design, such as... Figure 3 As shown, in terms of height design, the total tunnel height is calculated to be 4.15m using the formula H=h1+h2+h3+h4, corresponding to the track height (0.65m), lifting height (1.0m), transport equipment height (2.2m), and ground safety clearance (0.3m). Regarding width design, the width of a single-track tunnel is B=b1+b2+b3=0.85+1.75+1=3.6m; in the design of a double-track tunnel, considering greater transport demands and the safety clearance for oncoming traffic, the width is set to be no less than 6m. However, the single-track crane's operating path is fixed, resulting in poor flexibility.

[0005] Therefore, there is an urgent need for a turning path design method that can adapt to dynamic and complex environments and has high flexibility. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, this invention provides an autonomous obstacle avoidance and turning trajectory planning method for monorail cranes. This method involves acquiring distance information of the 360-degree environment surrounding the monorail crane using a sensor system installed on the crane, constructing a two-dimensional environment model, perceiving obstacles in the complex environment, and proposing a path planning algorithm. The algorithm reconstructs the trajectory based on the monorail crane's speed, acceleration, and turning radius. Finally, it constructs an objective function to find the optimal turning path that meets all kinematic constraints, thereby solving the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: An autonomous obstacle avoidance and turning trajectory planning method for monorail cranes includes the following steps: S1, The sensor system installed on the monorail can detect the distance data of the surrounding environment of the monorail equipment in real time; S2, based on distance data of the environment surrounding the monorail equipment, generates a two-dimensional model of the environment surrounding the monorail equipment through data fusion technology; S3. Calculate the optimal turning path of the monorail based on a two-dimensional model of the environment surrounding the monorail equipment. In step S3, a path planning algorithm is proposed to enable the monorail crane to autonomously plan its path in a mining environment and calculate the optimal turning path for the monorail crane, including the following steps: S31. In areas with dense obstacles, path search will generate non-optimal turning points. Therefore, path search is optimized by fusing collision detection mechanism. S32, a new obstacle appears, and the local path is dynamically replanned; S33 introduces a redundant node deletion mechanism after path planning to optimize path compression and avoid generating some unnecessary intermediate nodes or corner points during the node search process. S34: After the final path is generated, based on the speed, acceleration and turning radius constraints of the monorail equipment, the turning path is reconstructed using a fifth-order polynomial to calculate the optimal trajectory. By constructing an objective function that includes acceleration, curvature and speed, and using a particle swarm optimization algorithm to iteratively solve the problem, it is ensured that each path segment meets the kinematic constraints, and the globally optimal, continuous, smooth and controllable turning path is output.

[0008] As a further aspect of the present invention, in step S34, firstly, a preliminary analysis of the generated path is performed to identify regions with significant directional abrupt changes. For each turning segment with a directional abrupt change, the starting point, turning point, and ending point are extracted as key location points connecting two segments of quintic polynomial trajectories. For each curve segment (starting point → turning point → ending point), a quintic polynomial trajectory function of position with respect to time in the X-axis direction and a quintic polynomial trajectory function of position with respect to time in the Y-axis direction are constructed respectively, using the following method: 1. Determine six boundary conditions, including the initial position, target position, initial velocity, target velocity, initial acceleration, and target acceleration for each segment; 2. The initial velocity and initial acceleration are introduced from one segment of the path, and the target velocity and target acceleration are connected by the next segment to ensure continuity; 3. By modeling the state of the starting point and ending point in the spatial and temporal dimensions, the coefficients of each term are solved using a system of linear equations; the trajectory functions in both directions use time as the independent variable, and are accumulated sequentially through terms of different orders. After calculating the fifth-degree polynomial for each segment, the velocity function, acceleration function, and turning radius of the trajectory are obtained respectively. Based on the trajectory function of the fifth-degree polynomial for each segment, the rate of change of the path at any time is obtained, that is, the velocity of the path in the vertical direction. Then, the velocity is differentiated again to obtain the acceleration of the path at that time.

[0009] As a further aspect of the present invention, in step S34, based on this, a path curvature calculation expression is constructed to calculate the turning radius. The path curvature calculation expression obtains the velocity change of the path in the vertical direction per unit time by performing a first derivative operation on the path function. Subsequently, the second derivative result of the first derivative is obtained to obtain the acceleration in the vertical direction. The numerator of the path curvature calculation expression uses the square of the velocity change and a constant term to measure the "directional stability" of the path at that point. In its denominator, the acceleration of the path is used as an influencing factor to reflect the bending strength of the path. Subsequently, the velocity, acceleration, and turning radius are compared segment by segment with the kinematic constraints of the monorail equipment: 1. Whether the velocity exceeds the maximum travel speed at any point in time; 2. Whether the acceleration exceeds the maximum allowable acceleration; 3. Whether the turning radius is less than the minimum turning radius that the monorail equipment can withstand. If any kinematic constraint is not satisfied, the current path segment is determined to be an infeasible trajectory segment.

[0010] As a further aspect of the present invention, in step S34, if a segment of a fifth-order polynomial path is detected during trajectory optimization that does not satisfy the aforementioned kinematic constraints, an optimization adjustment phase is initiated. An objective function is constructed to find the optimal turning path that satisfies all kinematic constraints. The objective function introduces two variables: acceleration, path curvature, and the speed of the monorail equipment, and assigns corresponding weighting coefficients to adjust the influence of each variable during the optimization process. The objective function first weights the absolute value of the acceleration to reflect the dynamic stability of the path segment during execution. Then, it weights the absolute value of the reciprocal of the path curvature to measure the compactness of the turning radius; that is, the smaller the turning radius, the larger the curvature value, and the higher the cost. Finally, the speed of the monorail equipment on the path is introduced into the objective function and multiplied by the corresponding weighting coefficient to control the balance between time and energy consumption.

[0011] As a further aspect of the present invention, in step S34, the objective function is obtained by integrating the sum of the above two terms over the entire trajectory duration to obtain the total cost of a path within a given time period. A particle swarm optimization algorithm is used to solve the objective function, outputting the optimal turning path. Each particle's code includes the duration of the path segment, the fine-tuning of the control point position, and the boundary velocity direction angle. In each iteration, a corresponding fifth-degree polynomial trajectory is generated for each particle. It is determined whether the trajectory satisfies preset kinematic constraints such as velocity, acceleration, and minimum turning radius, and the objective function value is calculated. If the kinematic constraints are not met, the particle is considered an infeasible solution and does not participate in the update process of the local and global optimal solutions in the current round. If the kinematic constraints are met, the particle participates in the update process of the local and global optimal solutions until the algorithm converges or reaches the preset number of iterations. Each optimized path function is concatenated to construct a complete path. Between segments, the velocity and acceleration at the boundary are further continuously corrected to ensure that the entire path has no jumps in actual control. Finally, the globally optimal turning path is output, along with the parameter combination and performance indicators (path length, smoothness, and time) corresponding to the optimal turning path.

[0012] As a further aspect of the present invention, S1, a sensor system installed on the monorail gantry detects distance data of the surrounding environment in real time, including the following specific content: To obtain distance information of the 360-degree environment around the monorail gantry in real time, a sensor system is installed on the monorail gantry. The sensor system includes a lidar, an ultrasonic sensor, and an infrared sensor. The lidar, ultrasonic sensor, and infrared sensor work together to meet the perception requirements under different distances and environmental conditions. The lidar has high precision, high resolution, and long-range detection capabilities, and can quickly generate laser point cloud data of surrounding obstacles through laser beam scanning, achieving centimeter-level spatial modeling. The ultrasonic sensor is mainly used for mid-to-close range detection tasks, acquiring distance pulse data of mid-to-close range obstacles. The infrared sensor is used for near-range blind spot filling and obstacle identification in environments with poor visibility such as nighttime and dusty conditions, acquiring infrared thermal signals of surrounding obstacles.

[0013] As a further aspect of the present invention, S2, based on the distance data of the environment surrounding the monorail equipment, a two-dimensional model of the environment surrounding the monorail equipment is generated through data fusion technology, including the following specific content: the raw distance data collected by the lidar, ultrasonic sensor, and infrared sensor in the sensor system is transmitted to the central processing unit in real time through a high-speed communication interface. The central processing unit processes the raw distance data, including the following steps: S21, Time synchronization processing is performed on the distance data from the lidar, ultrasonic sensor and infrared sensor: The distance data is aligned with a unified timestamp according to the unified time base control module inside the central processing unit; The unified time base control module is composed of a hardware real-time clock chip and generates a stable time base signal through a crystal oscillator.

[0014] S22. Different sensors are installed at different positions on the monorail crane equipment, and the coordinate systems in which the data collected by each sensor are different. Therefore, spatial registration is required to unify the observation results of all sensors to a single body coordinate system. The translation vector and planar rotation matrix of each sensor relative to the body reference coordinate system are obtained by using the SLAM registration method and stored in the sensor parameter table. The coordinates of the laser point cloud, the ranging direction of the ultrasonic wave, and the position of the infrared heat source in each frame are uniformly converted into two-dimensional spatial points in the main coordinate system of the monorail crane equipment through Euler transformation.

[0015] S23 employs a confidence-weighted Bayesian fusion strategy, dynamically assigning weights based on the characteristics of different sensors to generate a two-dimensional model of the environment surrounding the monorail equipment in real time.

[0016] The technical effects and advantages of this invention, a method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes, are as follows: This invention integrates multimodal sensing units such as lidar, ultrasonic, and infrared sensors into the monorail crane to construct a two-dimensional environment model, achieving comprehensive perception and dynamic modeling of obstacles in complex environments. Based on this, a combination of SLAM spatial registration and Bayesian fusion strategies ensures the spatiotemporal consistency and semantic accuracy of the environmental perception data. In the path planning stage, a multi-strategy collaborative mechanism is introduced, including collision detection optimization, dynamic replanning, and redundant node compression, effectively improving path smoothness and obstacle avoidance robustness, avoiding path cornering and unnecessary backtracking. In trajectory generation, a fifth-order polynomial trajectory function is constructed, comprehensively considering kinematic constraints such as velocity, acceleration, and turning radius. The optimal path solution is iteratively solved using a particle swarm optimization algorithm, ensuring that the trajectory is continuous, differentiable, and controllable, meeting the physical performance requirements of the monorail crane during turning. This invention enables monorail cranes to autonomously, safely, and smoothly complete optimal path turns in dynamic and complex environments, improving their operational efficiency and environmental adaptability. Attached Figure Description

[0017] Figure 1 This is a flowchart of an autonomous obstacle avoidance and turning trajectory planning method for monorail crane equipment according to the present invention.

[0018] Figure 2 This is a structural diagram of a speed-adaptive monorail transport machine based on existing technology.

[0019] Figure 3 This is a schematic diagram of the cross-section design of a monorail operating tunnel in the existing technology.

[0020] Figure 4 This is a schematic diagram illustrating the deletion of redundant nodes according to the present invention.

[0021] Figure 5 This is a schematic diagram illustrating the optimization effect of redundant node removal in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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. Example 1

[0023] See Figure 1 The flowchart shown illustrates an autonomous obstacle avoidance and turning trajectory planning method for monorail cranes, comprising the following steps: S1, a sensor system installed on the monorail to detect distance data of the surrounding environment in real time.

[0024] S2, based on distance data of the environment surrounding the monorail equipment, generates a two-dimensional model of the environment surrounding the monorail equipment through data fusion technology.

[0025] S3. Based on a two-dimensional model of the environment surrounding the monorail equipment, calculate the optimal turning path for the monorail.

[0026] Further, S1, the sensor system installed on the monorail hoist detects distance data of the surrounding environment in real time, including: to obtain distance information of the 360-degree environment around the monorail hoist in real time, a sensor system is installed on the monorail hoist. The sensor system includes a lidar, an ultrasonic sensor, and an infrared sensor. The lidar, ultrasonic sensor, and infrared sensor work together to meet the perception requirements under different distances and environmental conditions. The lidar has high precision, high resolution, and long-range detection capabilities, and can quickly generate laser point cloud data of surrounding obstacles through laser beam scanning, achieving centimeter-level spatial modeling. The ultrasonic sensor is mainly used for mid-to-close range detection tasks, acquiring distance pulse data of mid-to-close range obstacles, especially suitable for detecting low-reflectivity targets or unstructured obstacles (such as mine debris, water accumulation, soft packaging materials, etc.). The infrared sensor is used for near-range blind spot filling and obstacle identification in poor visibility environments such as nighttime and dusty conditions, acquiring infrared thermal signals of surrounding obstacles.

[0027] The lidar system includes a main radar, side radars, and a rear-view radar. The main radar is installed at the upper center of the front of the monorail gantry, and is used to scan the path and obstacles ahead at a long distance to acquire high-precision two-dimensional point cloud data. The side radars are installed on the front sides of the monorail gantry to expand the lateral field of view, especially during equipment operation, to detect side walls, side obstacles at turns, and the shape of the tunnel sides. The rear-view radar is installed above the rear of the monorail gantry to detect obstacles behind during reversing or turning. The ultrasonic sensors are installed on the lower sides of the front of the monorail gantry, the left and right middle parts of the body, and the rear sides. The sensors on the lower sides of the front of the monorail gantry are used to detect low obstacles or ground protrusions (such as cables, debris, and rail intersections) at close range, filling in blind spots in the lidar's field of view. The sensors on the left and right middle parts of the body are used to measure the lateral distance between the equipment and the tunnel walls in real time to prevent deviation or side collisions. The sensors on the rear sides of the equipment are used to detect obstacles on the rear edge of the equipment or gaps between the equipment and the walls. The infrared sensors are respectively installed at the lower front edge of the monorail crane head (obliquely installed), at the four corners of the top of the vehicle body (tilted downwards at 15~30°), and at the bottom of the vehicle body (vertically downwards). The sensors installed at the lower front edge of the head are used for rapid identification of obstacles at close range, especially in low light, dusty, or high humidity environments, to help determine the outline of obstacles or temperature differences. The sensors installed at the four corners of the top of the vehicle body are used to identify near-field heat sources (such as workers, other equipment) or environmental temperature differences (such as ventilation openings, heat-generating equipment) around the monorail crane. The sensors installed at the bottom of the vehicle body are used to detect whether there is a break, displacement, or abnormal high temperature on the track (such as overheated cables), or to determine the alignment of the track position with the infrared markings.

[0028] Further, S2, based on the distance data of the environment surrounding the monorail equipment, a two-dimensional model of the environment surrounding the monorail equipment is generated through data fusion technology. This model includes: raw distance data collected by the lidar, ultrasonic sensor, and infrared sensor in the sensor system, which is transmitted in real time to the central processing unit via a high-speed communication interface (e.g., CAN bus, RS485, or Ethernet). The central processing unit processes the raw distance data, including the following steps: S21, Time synchronization processing is performed on the distance data from the lidar, ultrasonic sensor and infrared sensor: The distance data is aligned with a unified timestamp according to the unified time base control module inside the central processing unit; The unified time base control module is composed of a hardware real-time clock chip and generates a stable time base signal through a crystal oscillator.

[0029] S22. Different sensors are installed at different positions on the monorail crane equipment, and the coordinate systems in which the data collected by each sensor are different. Therefore, spatial registration is required to unify the observation results of all sensors to a single body coordinate system. The translation vector and planar rotation matrix of each sensor relative to the body reference coordinate system are obtained by using the SLAM registration method and stored in the sensor parameter table. The coordinates of the laser point cloud, the ranging direction of the ultrasonic wave, and the position of the infrared heat source in each frame are uniformly converted into two-dimensional spatial points in the main coordinate system of the monorail crane equipment through Euler transformation.

[0030] S23 employs a confidence-weighted Bayesian fusion strategy, dynamically assigning weights based on the characteristics of different sensors to generate a real-time two-dimensional model of the environment surrounding the monorail crane. Laser and lightning point clouds, serving as the backbone data source for high-precision geometric information, are given the highest confidence weight due to their high spatial resolution, high point cloud density, and strong structural reconstruction capabilities, forming the spatial framework of the two-dimensional environmental model. When a ranging hole is detected in a spatial region of the laser point cloud, and an ultrasonic wave returns an effective distance pulse at that location, the spatial region is marked as a "high-confidence obstacle point." The infrared thermal signal, combined with its spatial coordinates, assigns the semantic label "dynamic thermal body" to the point cloud of that spatial region. All fused data is projected and encoded into a unified two-dimensional spatial representation structure, managed and stored using a voxel grid. Each voxel unit not only records the distance value of the spatial location (obtained by weighted average of multi-source data), but also includes point cloud density (used to reflect surface morphology features), confidence distribution (showing the contribution ratio of radar, ultrasound, and infrared data), dynamic attribute labels (judged based on the difference between thermal signals and continuous frame point clouds), and semantic state of whether it is passable (based on the probability of obstacle occupancy).

[0031] Further, S3, based on the two-dimensional model of the environment surrounding the monorail equipment, the optimal turning path of the monorail is calculated, including the following specific content: Based on the two-dimensional model, a path planning algorithm is proposed to realize the autonomous operation path planning of the monorail equipment in a complex mine environment, and to calculate the optimal turning path of the monorail, including the following steps: S31. In areas with dense obstacles, path search may generate non-optimal turning points or even cause the path to "cross corners," meaning the search path passes through the edge or vertex of an obstacle. Therefore, a collision detection mechanism is used to optimize path search: When traversing the adjacent nodes s' of the current node s, a "connection validity judgment" is first performed, that is, to determine whether the path segment formed from s to s' intersects with the obstacles in the two-dimensional model. For the path segment formed by two non-adjacent nodes s and s', according to the definition of a rectangle in the two-dimensional grid coordinate system, the two diagonal points s1 and s2 of its corresponding matrix are calculated. If either of the points is in the obstacle area, it is initially determined that the path intersects with the obstacle, thus triggering an obstacle avoidance action. Conversely, if no intersection with an obstacle is detected in the path segment, the standard path cost calculation is performed. The search priority of the node is evaluated by adding the actual cost from the starting point to the current node and the heuristic estimate of the cost from the current node to the destination using Euclidean distance.

[0032] S32, When the environment changes dynamically, i.e., when new obstacles appear, the local path is dynamically replanned: When the front-end sensor detects a new obstacle, closed area or other dynamic change event, the two-dimensional model first marks the area that has changed in the voxel grid. Then, with the new obstacle point as the center, it reconstructs its set of adjacent nodes and identifies the affected path segments in the current path. For the nodes in the path segments, it recalculates the true cost from the starting point to the target point and the heuristic estimate from the node to the target point, updates its total cost function, and inserts the updated node into the priority queue.

[0033] S33 introduces a redundant node removal mechanism after path planning to optimize path compression and avoid generating unnecessary intermediate nodes or turning points during node search: such as Figure 4 As shown, an initial path is first generated, consisting of several consecutive nodes, with bends and unnecessary intermediate nodes along the path; then, for any pair of nodes in the entire path sequence... and The process involves traversing the path and attempting to connect the nodes with straight segments to construct a "transition path." Then, a collision mechanism is invoked to determine if the straight segment intersects any obstacle area. If the straight segment does not intersect any obstacle, all intermediate nodes are considered "redundant nodes" and can be safely deleted. and Establish connections, shorten paths, and reduce path complexity; if the straight line segment intersects with an obstacle boundary or vertex, mark the straight line segment as a "dangerous path" and terminate the current compression operation. Set this as the starting node for the next stage of path compression, and continue to perform the same judgment and processing on subsequent node sequences. Repeat the above process until the end of the path, then concatenate all valid straight line segments to form the final path. Figure 5 As shown, the original path has a large number of corner nodes, and the overall path is jagged. Each movement can only be in one of eight directions, so intermediate nodes are needed at multiple diagonal or turning points. The compressed path applies a "redundant node deletion mechanism", which attempts to connect distant nodes with straight lines in the path, and only retains the necessary turning points when straight lines cannot be used. The optimization result is a reduction in the number of nodes, a straighter and smoother path, and fewer turning points.

[0034] S34. After the final path is generated, the trajectory is reconstructed based on the speed, acceleration, and turning radius of the monorail equipment to calculate the optimal turning path. First, a preliminary analysis of the generated path is performed to identify regions with significant directional abrupt changes (i.e., locations where the path turns). For example, from node A→B→C, where vectors AB and BC are not collinear, this region is a turning segment. For each turning segment with a directional abrupt change, the starting point, turning point, and ending point are extracted as key location points connecting two segments of a fifth-order polynomial trajectory. For each curve segment (starting point → turning point → ending point), a fifth-order polynomial trajectory function of position with respect to time in the X-axis direction and a fifth-order polynomial trajectory function of position with respect to time in the Y-axis direction are constructed respectively. The specific method is as follows: 1. Determine six boundary conditions, including the initial position, target position, initial velocity, target velocity, initial acceleration, and target acceleration for each segment; 2. The initial velocity and initial acceleration are introduced from one segment of the path, and the target velocity and target acceleration are connected by the next segment to ensure continuity; 3. By modeling the state of the starting point and the ending point in the spatial and temporal dimensions, the coefficients of each term are solved using a system of linear equations; The trajectory functions in the above two directions all use time as the independent variable and are accumulated sequentially through terms of different orders. After calculating each segment of the fifth-order polynomial, the velocity function, acceleration function, and turning radius of the trajectory are obtained respectively. Based on the trajectory function of each segment of the fifth-order polynomial, the rate of change of the path's position at any given time is obtained, i.e., the velocity of the path in the vertical direction. Then, the velocity is differentiated once to obtain the acceleration of the path at that time. Based on this, a path curvature calculation expression is constructed to calculate the turning radius. The path curvature calculation expression obtains the change of the path's velocity in the vertical direction per unit time by performing a first-order derivative operation on the path function. Then, the second-order derivative result is obtained to obtain the acceleration in the vertical direction. The numerator of the path curvature calculation expression uses the square of the velocity change and a constant term to measure the "directional stability" of the path at that point. In its denominator, the acceleration of the path is used as an influencing factor to reflect the curvature strength of the path. Subsequently, the speed, acceleration, and turning radius are compared segment by segment with the kinematic constraints of the monorail equipment: 1. Whether the speed exceeds the maximum travel speed at any point in time; 2. Whether the acceleration exceeds the maximum allowable acceleration; 3. Whether the turning radius is less than the minimum turning radius that the monorail equipment can withstand; if any kinematic constraint is not met, the current path segment is determined to be an infeasible trajectory segment.

[0035] During trajectory optimization, if a segment of a fifth-order polynomial path is detected as not satisfying the aforementioned kinematic constraints, the optimization adjustment phase begins. An objective function is constructed to find the optimal turning path that meets all kinematic constraints. This objective function introduces two variables: acceleration, path curvature, and the speed of the monorail equipment, each with corresponding weighting coefficients to adjust their influence during optimization. First, the objective function weights the absolute value of the acceleration to reflect the dynamic stability of the path segment during execution. Then, it weights the absolute value of the reciprocal of the path curvature to measure the tightness of the turns; a smaller turning radius corresponds to a larger curvature value and higher cost. Finally, the speed of the monorail equipment on the path is incorporated into the objective function and multiplied by the corresponding weighting coefficient to control the balance between time and energy consumption. The objective function integrates the sum of these two variables over the entire trajectory duration to obtain the total cost of a path within a given time period. The objective function is solved using a particle swarm optimization algorithm, outputting the optimal turning path. Each particle's code includes the duration of the path segment, the fine-tuning of the control point position, and the boundary velocity direction angle. In each iteration, a corresponding fifth-degree polynomial trajectory is generated for each particle. It is determined whether the trajectory satisfies preset kinematic constraints such as velocity, acceleration, and minimum turning radius, and the objective function value is calculated. If the kinematic constraints are not met, the particle is considered an infeasible solution and does not participate in the update process of the local and global optima in the current iteration. If the kinematic constraints are met, the particle participates in the update process of the local and global optima until the algorithm converges or reaches the preset number of iterations. Each optimized path function segment is concatenated to construct a complete path. Between segments, the velocity and acceleration at the boundaries are further continuously corrected to ensure that the entire path has no jumps during actual control. Finally, the globally optimal turning path is output, along with the corresponding parameter combination and performance indicators (path length, smoothness, and time consumption).

[0036] This invention integrates multimodal sensing units such as LiDAR, ultrasonic, and infrared sensors into a monorail crane to construct a two-dimensional environmental model, enabling comprehensive perception and dynamic modeling of obstacles in complex environments. Based on this, it combines SLAM spatial registration and Bayesian fusion strategies to ensure the spatiotemporal consistency and semantic accuracy of environmental perception data. In the path planning stage, a multi-strategy collaborative mechanism is introduced, including collision detection optimization, dynamic replanning, and redundant node compression, effectively improving path smoothness and obstacle avoidance robustness, and preventing path crossings and unnecessary backtracking. For trajectory generation, a fifth-order polynomial trajectory function is constructed, comprehensively considering kinematic constraints such as velocity, acceleration, and turning radius. The optimal path solution is iteratively solved using a particle swarm optimization algorithm, ensuring that the trajectory is continuous, differentiable, and controllable, meeting the physical performance requirements of the monorail crane during turning. This invention enables the monorail crane to autonomously, safely, and smoothly complete optimal path turns in dynamic and complex environments, improving its operational efficiency and environmental adaptability.

[0037] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0038] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes, characterized in that, Includes the following steps: S1, The sensor system installed on the monorail can detect the distance data of the surrounding environment of the monorail equipment in real time; S2, based on distance data of the environment surrounding the monorail equipment, generates a two-dimensional model of the environment surrounding the monorail equipment through data fusion technology; S3. Calculate the optimal turning path of the monorail based on a two-dimensional model of the environment surrounding the monorail equipment. In step S3, a path planning algorithm is proposed to enable the monorail crane to autonomously plan its path in a mining environment and calculate the optimal turning path for the monorail crane, including the following steps: S31. In areas with dense obstacles, path search will generate non-optimal turning points. Therefore, path search is optimized by fusing collision detection mechanism. S32, a new obstacle appears, and the local path is dynamically replanned; S33 introduces a redundant node deletion mechanism after path planning to optimize path compression and avoid generating some unnecessary intermediate nodes or corner points during the node search process. S34: After the final path is generated, based on the speed, acceleration and turning radius constraints of the monorail equipment, the turning path is reconstructed using a fifth-order polynomial to calculate the optimal trajectory. By constructing an objective function that includes acceleration, curvature and speed, and using a particle swarm optimization algorithm to iteratively solve the problem, it is ensured that each path segment meets the kinematic constraints, and the globally optimal, continuous, smooth and controllable turning path is output.

2. The method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes according to claim 1, characterized in that... In step S34, a preliminary analysis of the generated path is performed to identify regions with obvious abrupt changes in direction. The starting point, turning point, and ending point are extracted as key location points connecting the two segments of the fifth-order polynomial trajectory. For each segment of the curve, a fifth-order polynomial trajectory function of position with respect to time in the X-axis direction and a fifth-order polynomial trajectory function of position with respect to time in the Y-axis direction are constructed respectively. After calculating the fifth-order polynomial for each segment, the velocity function, acceleration function, and turning radius of the trajectory are obtained respectively. Based on the fifth-order polynomial trajectory function of each segment, the rate of change of the path position at any time is obtained. Then, the velocity is differentiated once to obtain the acceleration of the path at that time.

3. The method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes according to claim 1, characterized in that... In step S34, a path curvature calculation expression is constructed to calculate the turning radius. This expression obtains the velocity change of the path in the vertical direction per unit time by performing a first derivative operation on the path function. Then, a second derivative is performed on the first derivative result to obtain the acceleration in the vertical direction. The numerator of the path curvature calculation expression is expressed using the square of the velocity change and a constant term. In its denominator, the path acceleration is used as an influencing factor to reflect the bending strength of the path. The velocity, acceleration, and turning radius are compared segment by segment with the kinematic constraints of the monorail equipment.

4. The method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes according to claim 1, characterized in that, In step S34, during the trajectory optimization process, if a certain segment of a fifth-order polynomial path is detected as not satisfying the above kinematic constraints, the optimization and adjustment phase begins. An objective function is constructed to find the optimal turning path that satisfies all kinematic constraints. The objective function introduces two variables: acceleration, path curvature, and monorail equipment travel speed, and sets corresponding weight coefficients to adjust the influence of each variable in the optimization process. The objective function first weights the absolute value of the acceleration to reflect the dynamic stability of the path segment during execution. Subsequently, the inverse of the path curvature is weighted by absolute value to measure the tightness of the path's turns; finally, the travel speed of the monorail equipment on the path is introduced into the objective function and multiplied by the corresponding weight coefficient to control the balance between time and energy consumption of the path.

5. The method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes according to claim 1, characterized in that, In step S34, a particle swarm optimization algorithm is used to solve the objective function and output the optimal turning path. Each particle's code includes the duration of the path segment, the fine-tuning amount of the control point position, and the boundary velocity direction angle. In each iteration, a corresponding fifth-order polynomial trajectory is generated for each particle, and its kinematic constraints are judged. The objective function value is calculated. If the kinematic constraints are not met, the particle is considered an infeasible solution and does not participate in the update process of the local and global optimal solutions in the current round. If the kinematic constraints are met, the particle participates in the update process of the local and global optimal solutions until the algorithm converges or reaches the preset number of iterations. Each optimized path function segment is spliced ​​together to construct a complete path. Between segments, the velocity and acceleration at the boundary are further continuously corrected to ensure that the entire path has no jumps in actual control. Finally, the globally optimal turning path is output.

6. The method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes according to claim 1, characterized in that, The sensor system in S1 includes a lidar, an ultrasonic sensor, and an infrared sensor; the lidar is used for long-range detection and generating laser point cloud data; the ultrasonic sensor is used for mid-to-short-range detection of unstructured obstacles; and the infrared sensor is used for short-range blind spot detection and obstacle identification in low-visibility environments.

7. The method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes according to claim 1, characterized in that, The lidar system includes a main radar, side radars, and a rear-view radar. The main radar is installed at the upper center of the front of the monorail gantry crane vehicle to scan the path and obstacles ahead. The side radars are installed on the front sides of the vehicle body to detect side walls and side obstacles at turns. The rear-view radar is installed above the rear of the vehicle to detect obstacles behind when reversing or turning. The ultrasonic sensors are installed on the lower sides of the front of the vehicle, the left and right middle parts of the vehicle body, and the rear sides. The sensors on the lower sides of the front of the vehicle are used to detect low obstacles at close range ahead. The sensors on the left and right middle parts of the vehicle body are used to measure the lateral distance to the tunnel side walls. The sensors on the rear sides of the vehicle body are used to detect obstacles on the rear edge. The infrared sensors are installed on the lower front edge of the front of the vehicle, the four corners of the top of the vehicle body, and the bottom of the vehicle body. The sensor on the lower front edge of the front of the vehicle is used for near-range obstacle identification. The sensors at the four corners of the top are used to identify near-field heat sources. The sensors at the bottom are used to detect track status.

8. The method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes according to claim 1, characterized in that, The data fusion technology in S2 includes the following steps: S21, the sensor data is synchronized in time and a unified timestamp is achieved through the unified time base control module of the central processing unit; S22, the SLAM registration method is used to obtain the translation vector and rotation matrix of each sensor relative to the machine coordinate system, and all data are transformed to the main coordinate system of the monorail crane equipment through Euler transformation; S23, based on a confidence-weighted Bayesian fusion strategy, uses laser point cloud as the skeleton, fuses ultrasonic and infrared data, and generates a two-dimensional model in the form of voxel grid. The voxel unit records distance value, point cloud density, confidence distribution, dynamic attribute label and traffic semantic state.

9. The method for autonomous obstacle avoidance and turning trajectory planning for monorail cranes according to claim 1, characterized in that, In S23, the laser point cloud is assigned the highest confidence weight, the valid data returned by the ultrasonic wave in the void region of the laser point cloud is marked as "high confidence obstacle point", and the infrared thermal signal adds the semantic label of "dynamic thermal body" to the point cloud.

Images