A cloud-based hybrid local path planning method

Abstract

This invention discloses a cloud-controlled hybrid local path planning method, comprising: an unmanned mining truck system acquiring a reference line for travel and generating a local path based on an RTK local planner; triggering forward obstacle avoidance when an obstacle is encountered; the unmanned mining truck system acquiring surrounding environmental information in real time and feeding it back to the cloud control; when the cloud control determines that the obstacle avoidance conditions are met, issuing an obstacle avoidance command; if the obstacle avoidance conditions are not met, maintaining forward obstacle avoidance and waiting for manual intervention; Motion calling the obstacle avoidance local planner to generate an obstacle avoidance local path until obstacle avoidance is completed; after obstacle avoidance, the mining truck returns to the predetermined travel path, calls the original RTK local planner, and continues traveling. This invention, with the assistance of cloud control, improves operational efficiency while ensuring safety, and simultaneously achieves the hybrid use of two local path planning methods, which can flexibly adapt to different environments and save computing resources.

Description

Technical Field

[0001] This invention relates to the field of unmanned mining truck technology, and in particular to a cloud-based hybrid local path planning method. Background Technology

[0002] In the current field of unmanned mining trucks, due to the complex road conditions in mines, the perception module alone cannot identify all potholes, water accumulations, and other scenarios. Blindly avoiding obstacles can easily cause damage to vehicles and equipment. When multiple mining trucks operate simultaneously in a mine, there is no coordination between them, which can easily cause collisions and blockages during obstacle avoidance, affecting normal mine production. A single obstacle avoidance plan requires continuous iterative calculations, which consumes a lot of computing resources, requires high computing power, and lacks fault tolerance. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the existing technology. To achieve the above objective, a cloud-based hybrid local path planning method is adopted to solve the problems mentioned in the background technology.

[0004] A cloud-based hybrid local path planning method includes the following steps:

[0005] Step S1: The unmanned mining truck system obtains the reference line for driving and generates a local path based on the RTK local planner. When an obstacle appears, it triggers the forward obstacle-stopping perception.

[0006] Step S2: The unmanned mining truck system acquires surrounding environmental information in real time and feeds it back to the cloud control. When the cloud control determines that the obstacle avoidance conditions are met, it issues an obstacle avoidance command. If the obstacle avoidance conditions are not met, it maintains forward perception and stops waiting for manual intervention.

[0007] Step S3: Motion calls the obstacle avoidance local planner to generate a local obstacle avoidance path until the obstacle avoidance is completed.

[0008] Step S4: After the obstacle avoidance is completed, the minecart returns to the predetermined driving path, calls the original RTK local planner, and continues driving.

[0009] As a further aspect of the present invention, the specific steps in step S1 include:

[0010] Based on the coordinates of the projection point of the mining truck on the reference line at each moment, a series of local path points are generated by sampling from the reference line, where the length of the local path needs to be configured.

[0011] Within dense sampling intervals, a smaller sampling step size is set to generate path points at a higher density;

[0012] Within the sparse sampling interval, the sampling step size is set to be relatively large, so that path points are generated at a lower density.

[0013] The area closer to the mine car projection point is the dense sampling area, while the area farther away is the sparse sampling area.

[0014] As a further aspect of the present invention: when an obstacle appears in step S1, the specific steps for triggering the forward obstacle sensing and stopping include:

[0015] When an obstacle appears in front that affects the normal movement of the mine truck, the mine truck begins to slow down through the sensing information until it stops;

[0016] Trigger forward obstacle stopping by sensing, where the local path length planned by the RTK local planner is 0.

[0017] As a further aspect of the present invention, the specific steps in step S2 include:

[0018] The mining trucks upload information about their surrounding environment and the passage of nearby mining trucks to the cloud control system via a panoramic system.

[0019] Once the cloud control confirms that the safety conditions are met, it issues an obstacle avoidance command and switches to the obstacle avoidance local planner; otherwise, it maintains the forward perception obstacle-stopping state.

[0020] As a further aspect of the present invention, the specific steps in step S3 include:

[0021] Step S31: When switching to the obstacle avoidance local planner, a local path to avoid obstacles is planned in real time for each cycle;

[0022] The current position coordinates (x0, y0) of the minecart are converted to (s0, l0) in the Frenet coordinate system according to the reference line; the widths of the left and right roads are obtained, and the left and right road boundaries are established, i.e., the sampling boundaries; sampling is performed uniformly according to the boundaries in each cycle, and n road points are sampled in each layer;

[0023] Step S32: Based on the sampling points of each layer and the vehicle's s, 1, l′, l″, where The smooth curve connects each sampling point of each layer, and the equation of each curve in the Frenet coordinate system can be obtained by using a fifth-order polynomial.

[0024] Step S33: Calculate the cost of each path. The cost of each path consists of three parts: smoothing cost, obstacle collision cost, and reference line guidance cost.

[0025] C total (f(s))=C smooth (f)+C obs (f)+C guidance (f)

[0026] Among them, C smooth (f) represents the path smoothing cost, C obs(f) represents the obstacle collision cost, C guidance (f) represents the smoothing cost of the reference line;

[0027]

[0028]

[0029] C guidance (f)=∫(f(s)-g(s)) 2 ds

[0030] In the formula, f(s) is the path, f`(s) is the path heading angle, f``(s) is the path curvature, f```(s) is the derivative of the path curvature, and d is the distance between the obstacle and the vehicle. c d is the safe distance threshold. n The threshold for slight avoidance distance is given by g(s), which is the equation of the reference line.

[0031] Step S34: Among all the smooth paths generated in step S32, select the total cost C. total The path with the smallest f(s) is taken as the local path for the final output of the current cycle;

[0032] When adc_end_s > obs_end_s + safe_threshold && l0 < 0.3, the system determines that obstacle avoidance has ended and switches to the RTK local planner to plan the local path. Here, adc_end_s is the s value of the minecart's tail in the Frenet coordinate system, obs_end_s is the maximum s value of the obstacle's sl_boundary in the Frenet coordinate system, and safe_threshold is the safe boundary value.

[0033] Compared with the prior art, the present invention has the following technical advantages:

[0034] The above technical solution generates a local path through an RTK local planner. When an obstacle is encountered, a forward obstacle-stopping mechanism is triggered. The mining truck feeds back the surrounding environment information to the cloud control system. When the cloud control system determines that the obstacle avoidance conditions are met, it issues an obstacle avoidance command. Motion calls the obstacle avoidance local planner to generate an obstacle avoidance local path until the obstacle avoidance is completed. Interaction with the cloud control system improves obstacle avoidance safety without affecting the normal operation of other mining trucks. The hybrid use of two local path planning methods allows the mining truck to flexibly adapt to different environments, ensuring reliable operation in complex mining environments. It can also significantly save computational resources, further facilitating the efficient operation of the unmanned driving system. Attached Figure Description

[0035] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings:

[0036] Figure 1 This is a schematic diagram illustrating the steps of the path planning method according to an embodiment of this application;

[0037] Figure 2 This is a flowchart illustrating the path planning method according to an embodiment of this application.

[0038] Figure 3 This is a schematic diagram illustrating the principle of local path generation in an embodiment of this application. Detailed Implementation

[0039] 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.

[0040] Please refer to Figure 1 and Figure 2 In this embodiment of the invention, a cloud-controlled hybrid local path planning method includes the following steps:

[0041] Step S1: The unmanned mining truck system acquires a reference line for travel and generates a local path based on the RTK local planner. When an obstacle is encountered, it triggers a forward obstacle-sensing obstacle-stopping mechanism. Specific steps include:

[0042] In this embodiment, the road centerline is smoothed to generate a smooth reference line, which is provided to the local planner as a reference to generate the final local path.

[0043] Based on the coordinates of the projection point of the mining truck on the reference line at each moment, a series of local path points are generated by sampling from the reference line, where the length of the local path needs to be configured.

[0044] Within dense sampling intervals, a smaller sampling step size is set to generate path points at a higher density;

[0045] Within the sparse sampling interval, the sampling step size is set to be relatively large, so that path points are generated at a lower density.

[0046] The area closer to the mine car projection point is the dense sampling area, while the area farther away is the sparse sampling area.

[0047] The specific steps for triggering forward obstacle stopping when an obstacle is encountered include:

[0048] When an obstacle appears in front that affects the normal movement of the mine truck, the mine truck begins to slow down through the sensing information until it stops;

[0049] Trigger forward obstacle stopping by sensing, where the local path length planned by the RTK local planner is 0.

[0050] Step S2: The unmanned mining truck system acquires real-time information about the surrounding environment and feeds it back to the cloud control system. When the cloud control system determines that the obstacle avoidance conditions are met, it issues an obstacle avoidance command. If the obstacle avoidance conditions are not met, the system maintains forward perception and stops, waiting for manual intervention. Specific steps include:

[0051] The mining trucks upload information about their surrounding environment and the passage of nearby mining trucks to the cloud control system via a panoramic system.

[0052] Once the cloud control confirms that the safety conditions are met, it issues an obstacle avoidance command and switches to the obstacle avoidance local planner; otherwise, it maintains the forward perception obstacle-stopping state.

[0053] Step S3: Motion calls the obstacle avoidance local planner to generate a local obstacle avoidance path until the obstacle avoidance is completed. Specific steps include:

[0054] Step S31: When switching to the obstacle avoidance local planner, a local path to avoid obstacles is planned in real time in each cycle. Due to the complexity and real-time requirements of the obstacle avoidance local planner, relatively large computing resources are necessary, so the computing resource consumption is large.

[0055] like Figure 3 As shown, the diagram illustrates the principle of local path generation.

[0056] The current position coordinates (x0, y0) of the minecart are converted to (s0, l0) in the Frenet coordinate system according to the reference line; the widths of the left and right roads are obtained, and the left and right road boundaries are established, i.e., the sampling boundaries; sampling is performed uniformly according to the boundaries in each cycle, and n road points are sampled in each layer;

[0057] Step S32: Based on the sampling points of each layer and the vehicle's s, 1, l′, l″, where The smooth curve connects each sampling point of each layer, and the equation of each curve in the Frenet coordinate system can be obtained by using a fifth-order polynomial.

[0058] Step S33: Calculate the cost of each path. The cost of each path consists of three parts: smoothing cost, obstacle collision cost, and reference line guidance cost.

[0059] C total (f(s))=C smooth (f)+C obs (f)+C guidance (f)

[0060] Among them, C smooth (f) represents the path smoothing cost, C obs(f) represents the obstacle collision cost, C guidance (f) represents the smoothing cost of the reference line;

[0061] C smooth (f)=w1∫(f′(s)) 2 ds+w2∫(f″(s)) 2 ds+w3∫(f″′s)) 2 ds.

[0062]

[0063] c guidance (f)=∫(f(s)-g(s)) 2 ds

[0064] In the formula, f(s) is the path, f`(s) is the path heading angle, f``(s) is the path curvature, f```(s) is the derivative of the path curvature, and d is the distance between the obstacle and the vehicle. c d is the safe distance threshold. n The threshold for slight avoidance distance is given by g(s), which is the equation of the reference line.

[0065] Step S34: Among all the smooth paths generated in step S32, select the total cost C. total The path with the smallest f(s) is taken as the local path for the final output of the current cycle;

[0066] When `adc_end_s > obs_end_s + safe_threshold && l0 < 0.3`, the system determines that obstacle avoidance has ended. At this point, it switches to the RTK local planner to plan the local path. Here, `adc_end_s` is the `s` value of the minecart's tail in the Frenet coordinate system, `obs_end_s` is the maximum `s` value of the obstacle's `sl_boundary` in the Frenet coordinate system, and `safe_threshold` is the safety domain.

[0067] Step S4: After the obstacle avoidance is completed, the minecart returns to the predetermined driving path, calls the original RTK local planner, and continues driving.

[0068] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention. The scope of the invention is defined by the appended claims and their equivalents, all of which should be included within the scope of protection of the invention.

Claims

1. A cloud-based hybrid local path planning method, characterized in that, Includes the following steps: Step S1: The unmanned mining truck system obtains the reference line for driving and generates a local path based on the RTK local planner. When an obstacle appears, it triggers the forward obstacle-stopping perception. Step S2: The unmanned mining truck system acquires surrounding environmental information in real time and feeds it back to the cloud control. When the cloud control determines that the obstacle avoidance conditions are met, it issues an obstacle avoidance command. If the obstacle avoidance conditions are not met, it maintains forward perception and stops at the obstacle, waiting for manual intervention. Step S3: Motion calls the obstacle avoidance local planner to generate a local obstacle avoidance path until the obstacle avoidance is completed. The specific steps include: Step S31: When switching to the obstacle avoidance local planner, a local path to avoid obstacles is planned in real time for each cycle; The current position coordinates of the minecart Convert to Frenet coordinate system based on reference lines Obtain the width of the left road and the width of the right road, and establish the left and right road boundaries, i.e., the sampling boundaries; sample uniformly according to the boundaries in each cycle, and sample n road points in each layer; Step S32: Based on the sampling points of each layer and the vehicle's s, l, , ,in = , The smooth curve connects each sampling point of each layer, and the equation of each curve in the Frenet coordinate system can be obtained by using a fifth-order polynomial. Step S33: Calculate the cost of each path. The cost of each path consists of three parts: smoothing cost, obstacle collision cost, and reference line guidance cost. in, For path smoothing cost, The cost of obstacle collisions, The cost of smoothing the reference line; In the formula, For path, The path heading angle, For path curvature, Let be the derivative of the path curvature, and d be the distance between the obstacle and the vehicle. For safe distance threshold, This is the threshold for slight avoidance distance. Equation for the reference line; Step S34: Select the total cost from all the smooth paths generated in step S32. The shortest path is used as the local path for the final output of the current cycle; When adc_end_s > obs_end_s + safe_threshold && At this point, the system determines that the obstacle avoidance is over and switches to the RTK local planner to plan the local path. Here, adc_end_s is the s value of the minecart tail in the Frenet coordinate system, obs_end_s is the maximum s value of the obstacle in the Frenet coordinate system sl_boundary, and safe_threshold is the safety threshold. Step S4: After the obstacle avoidance is completed, the minecart returns to the predetermined driving path, calls the original RTK local planner, and continues driving.

2. The cloud-based hybrid local path planning method according to claim 1, characterized in that, The specific steps in step S1 include: Based on the coordinates of the projection point of the mining truck on the reference line at each moment, a series of local path points are generated by sampling from the reference line, where the length of the local path needs to be configured.

3. The cloud-based hybrid local path planning method according to claim 1, characterized in that, When an obstacle appears in step S1, the specific steps for triggering forward obstacle detection and stopping include: When an obstacle appears in front that affects the normal movement of the mine truck, the mine truck begins to slow down through the sensing information until it stops; Trigger forward obstacle stopping by sensing, where the local path length planned by the RTK local planner is 0.

4. The cloud-based hybrid local path planning method according to claim 1, characterized in that, The specific steps in step S2 include: The mining trucks upload information about their surrounding environment and the passage of nearby mining trucks to the cloud control system via a panoramic system. Once the cloud control confirms that the safety conditions are met, it issues an obstacle avoidance command and switches to the obstacle avoidance local planner; otherwise, it maintains the forward perception obstacle-stopping state.

Images