A robot path planning method, device, equipment, medium and product

Abstract

The application discloses a robot path planning method, device, equipment, medium and product, and relates to the field of robot obstacle avoidance. The method comprises the following steps: constructing a grid map of a target environment; obtaining a global planning path by using an improved A* algorithm according to the grid map; constructing a dynamic heuristic function based on a geometric distance and a dynamic risk parameter on the basis of the A* algorithm; obtaining a plurality of local planning paths by using a path decomposition method and an improved dynamic window algorithm according to the global planning path; constructing an evaluation function based on a cross-modal fusion method and a collision time method on the basis of the dynamic window algorithm; and obtaining a robot path planning result according to all the local planning paths. The application realizes global path planning and local path planning of a robot based on the improved A* algorithm and the improved dynamic window algorithm, improves path planning efficiency and safety, and enhances the obstacle avoidance capability of the robot in a complex environment.

Description

Technical Field

[0001] This application relates to the field of robot obstacle avoidance, and in particular to a robot path planning method, apparatus, equipment, medium, and product. Background Technology

[0002] Global path planning for robots is primarily used to provide the robot with an overall route from its starting point to its destination. Traditional global path planning methods often struggle to respond quickly and adapt to changes in complex environments with constantly evolving obstacles.

[0003] Existing global path planning methods perform well in static environments, effectively finding the shortest path from the starting point to the target point. Static obstacles are typically fixed in the environment, and different types of static obstacles (such as equipment and pipes) require robots to adopt different obstacle avoidance strategies. However, when facing dynamically changing obstacles, the algorithms' response speed is slow, often requiring path recalculation. Furthermore, complex environments feature diverse obstacles and varying ground conditions, further increasing the difficulty of global path planning. Improving the stability and accuracy of algorithms in complex and dynamic environments remains a major challenge for global path planning. Therefore, improving the adaptability and computational efficiency of algorithms in complex and dynamic environments while ensuring the accuracy and stability of global path planning has become a core issue in current global path planning research. Summary of the Invention

[0004] The purpose of this application is to provide a robot path planning method, apparatus, device, medium, and product, which can realize global and local path planning for robots based on the improved A* algorithm and the improved dynamic window algorithm, thereby improving the efficiency and safety of path planning and enhancing the robot's obstacle avoidance ability in complex environments.

[0005] To achieve the above objectives, this application provides the following solution:

[0006] Firstly, this application provides a robot path planning method, including:

[0007] Construct a raster map of the target environment;

[0008] Based on the grid map, an improved A* algorithm is used to obtain a global planning path; the improved A* algorithm is based on the A* algorithm and constructs a dynamic heuristic function based on geometric distance and dynamic risk parameters.

[0009] Based on the global planning path, multiple local planning paths are obtained by using a path decomposition method and an improved dynamic window algorithm; the improved dynamic window algorithm constructs an evaluation function based on a cross-modal fusion method and a collision time method on the basis of the dynamic window algorithm.

[0010] Based on all the local planning paths, the robot path planning results are obtained.

[0011] Secondly, this application provides a robot path planning device, comprising:

[0012] The map building module is used to build a raster map of the target environment;

[0013] The global path acquisition module is used to obtain a globally planned path based on the grid map and an improved A* algorithm; the improved A* algorithm is based on the A* algorithm and constructs a dynamic heuristic function based on geometric distance and dynamic risk parameters.

[0014] The local path acquisition module is used to obtain multiple local planned paths based on the global planned path, using a path decomposition method and an improved dynamic window algorithm; the improved dynamic window algorithm constructs an evaluation function based on the cross-modal fusion method and the collision time method on the basis of the dynamic window algorithm;

[0015] The path planning module is used to obtain the robot path planning result based on all the local planned paths.

[0016] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the robot path planning method described above.

[0017] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the robot path planning method described above.

[0018] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the robot path planning method described above.

[0019] According to the specific embodiments provided in this application, the following technical effects are disclosed:

[0020] This application provides a robot path planning method, apparatus, device, medium, and product. It constructs a dynamic heuristic function for an improved A* algorithm based on geometric distance and dynamic risk parameters, using the improved A* algorithm as the basis for global path planning, significantly improving the efficiency and safety of path planning. Furthermore, it constructs an evaluation function for an improved dynamic window algorithm based on cross-modal fusion and collision time methods. By improving the dynamic window algorithm, it achieves differentiated obstacle avoidance strategies for different obstacle types, effectively suppressing path oscillations and enhancing the robustness of dynamic obstacle avoidance. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is an application environment diagram of a robot path planning method according to an embodiment of this application;

[0023] Figure 2 A flowchart illustrating a robot path planning method provided in an embodiment of this application;

[0024] Figure 3 This is a schematic diagram of the overall framework of a robot path planning method provided in an embodiment of this application;

[0025] Figure 4 This is a schematic diagram illustrating the speed relationship between a robot and a dynamic obstacle, provided in one embodiment of this application.

[0026] Figure 5 This is a schematic diagram of the functional modules of a robot path planning device provided in an embodiment of this application;

[0027] Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0028] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0029] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0030] The robot path planning method provided in this application embodiment can be applied to, for example, Figure 1In the application environment shown, terminal 101 communicates with server 102 via a network. A data storage system can store the data that server 102 needs to process. The data storage system can be set up independently, integrated into server 102, or placed in the cloud or on another server. Terminal 101 can send a grid map of the target environment to server 102. After receiving the grid map, server 102 uses an improved A* algorithm to obtain a global planning path based on the grid map. Then, based on the global planning path, it uses a path decomposition method and an improved Dynamic Window Approach (DWA) to obtain multiple local planning paths, thus obtaining the robot path planning result. Server 102 can feed back the obtained robot path planning result to terminal 101. Furthermore, in some embodiments, the robot path planning method can also be implemented independently by server 102 or terminal 101. For example, terminal 101 can directly process the grid map of the target environment to obtain the robot path planning result, or server 102 can obtain the grid map of the target environment from the data storage system and further process it to obtain the robot path planning result.

[0031] The terminal 101 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. The server 102 can be implemented using a standalone server or a server cluster composed of multiple servers, or it can be a cloud server.

[0032] In one exemplary embodiment, such as Figure 2 As shown, a robot path planning method is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 102 as an example, the explanation includes the following steps 201 to 204. Wherein:

[0033] Step 201: Construct a raster map of the target environment. The raster map can represent different areas of the environment in detail, providing basic environmental information for path planning.

[0034] Step 202: Based on the grid map, an improved A* algorithm is used to obtain a global planning path. The improved A* algorithm constructs a dynamic heuristic function based on geometric distance and dynamic risk parameters, building upon the A* algorithm. By comprehensively considering the dynamic heuristic function and cost function between the starting point and the target node, the improved A* algorithm calculates an optimal global planning path.

[0035] Step 203: Based on the global planning path, multiple local planning paths are obtained using a path decomposition method and an improved dynamic window algorithm. The improved dynamic window algorithm constructs an evaluation function based on a cross-modal fusion method and a collision time method, building upon the dynamic window algorithm.

[0036] Step 204: Obtain the robot path planning result based on all local planning paths.

[0037] In a preferred embodiment, after step 202 above, the robot path planning method further includes:

[0038] The global planning path is optimized by sequentially employing the artificial potential field method and the RDP algorithm.

[0039] In this embodiment, the real-time obstacle avoidance capability of the artificial potential field method is deeply integrated with the global optimality of the improved A* algorithm, which can achieve synergistic optimization of path safety and efficiency.

[0040] (1) Potential field construction

[0041] The potential field is composed of a superposition of gravitational and repulsive fields, guiding the robot to move towards the target node and avoid obstacles. The gravitational field is generated by the target node, and its strength decreases with distance, guiding the robot towards the target. Its calculation formula is as follows:

[0042] F att =k att ·(p goal -p current ).

[0043] Among them, F att Let k be the gravitational field strength. att p is the gravitational gain coefficient used to control the gravitational field strength. goal Let p be the coordinate vector of the target node. current This is the robot's current coordinate vector.

[0044] The repulsive field is generated by the obstacle, and its intensity decreases non-linearly with increasing distance, but significantly increases within a dynamic safety threshold. Its calculation formula is as follows:

[0045]

[0046] Among them, Frep Let k be the repulsive field strength. rep d represents the repulsive force gain coefficient, which is dynamically adjusted according to the obstacle threat level. safe For dynamic security thresholds, e obs Let d be the unit direction vector of the robot pointing towards the obstacle, d0 be the maximum effective range of the repulsive field, and d r The distance between the robot and the obstacle is expressed in Euclidean form.

[0047] (2) Path correction

[0048] Based on the globally planned path generated by the improved A* algorithm, path safety correction is achieved through the superposition of potential forces. For the globally planned path P = {p1, p2, ..., p...} generated by the improved A* algorithm... n Apply potential force to each node in the algorithm and calculate the offset as shown in the following formula. Iterate and adjust the global planning path to move it away from high-risk areas until the maximum number of iterations is reached or the node offset is less than a set threshold.

[0049] Δp i =ζ·(F att (p i )+∑F rep,n (p i )).

[0050] Where, Δp i For node p i The corresponding node offset, where ζ is the weighting coefficient of the resultant force field strength.

[0051] The artificial potential field method is a supplement to the dynamic heuristic function. Its purpose is to keep the globally planned path at a certain distance from obstacles, so as to ensure the safe distance of the robot when avoiding obstacles.

[0052] The RDP algorithm can reduce invalid polyline segments and improve path smoothness. Its core idea is to select a path {Q0, Q1, ..., Q...} containing several discrete points. m}, the first and last points Q0 and Q m As a reserved point. For each discrete point Q in the middle. i (i = 1, 2, ..., m-1), calculate the line Q0 connecting the beginning and end of the line (i = 1, 2, ..., m-1). m Find the vertical distance and record the point Q corresponding to the maximum distance. k The formula for calculating vertical distance is as follows:

[0053]

[0054] Where s is the midpoint Q i Connect the beginning and end Q0 Q mThe vertical distance, x1 and y1 are the x and y coordinates of the first point Q0, respectively, and x2 and y2 are the vertical distances of the tail point Q0. m The x and y coordinates are given, where x0 and y0 are the coordinates of the midpoint Q. i The horizontal and vertical coordinates.

[0055] If the calculated maximum vertical distance s max If the distance is less than a given threshold, the entire path can be completely approximated by the first and last points, eliminating the need to retain intermediate points. If the calculated maximum vertical distance s... max If the distance is greater than a given threshold, then the midpoint Q corresponding to the maximum vertical distance is... k The entire path is divided into Q0Q. k With Q k Q m Two segments are created, and the above steps are repeated for each of the two segmented paths.

[0056] In a preferred embodiment, the dynamic risk parameters include obstacle density, ground slip coefficient, and moving obstacle function.

[0057] The dynamic heuristic function of the improved A* algorithm is:

[0058] h(n) = α·h static (n)+β·h dynamic (n).

[0059]

[0060] f2(n)=SlipperyScore(n)·μ.

[0061]

[0062] Where h(n) is the dynamic heuristic function, h static (n) represents the static distance based on the raster map, h dynamic (n) is the dynamic adjustment term, and α is h static (n) corresponds to the adaptive weight parameter, where β is h dynamic (n) corresponds to the adaptive weight parameter, x n Let y be the x-coordinate of the nth node in the raster map. n Let x be the ordinate of the nth node in the raster map. goal Let y be the x-coordinate of the target node. goal Let γ(n) be the ordinate of the target node, γ(n) be the passage cost coefficient of the grid cell containing the nth node in the grid map, f1(n) be the obstacle density, and N be the distance between the nodes. obsf(n) represents the number of obstacles, π is pi, r is the statistical radius, λ is the normalization coefficient, f2(n) is the ground slip coefficient, SlipperyScore(n) is the slip score, μ is the slip cost weight, f3(n) is the moving obstacle function, max(·) is the maximum value function, and d min (n) is the minimum distance between the predicted trajectory and the nth node in the grid map, ∈ is the smoothing factor, w1 is the weight coefficient corresponding to f1(n), w2 is the weight coefficient corresponding to f2(n), and w3 is the weight coefficient corresponding to f3(n).

[0063] The A* algorithm initializes by loading a static map and generating an open list, then proceeds to the path planning phase. During path planning, the starting point is added to the open list, and its surrounding accessible neighboring nodes are expanded, adding these nodes to the open list as well. Simultaneously, the starting point is moved to the closed list, indicating it has been visited. Then, the node with the smallest heuristic function value in the open list is selected as the current expanded node, removed from the open list, and added to the closed list. The algorithm then continues searching for unvisited and unavailable reachable nodes not in the closed list, using this node as the center. When a neighboring node already exists in the open list, the algorithm calculates the cost of the new path formed by the current node through this neighboring node and compares it with the cost of the existing path. If the new path has a lower cost, the parent node of the current node is updated to the current node, and its cost is adjusted accordingly; otherwise, if the original path is better, it remains unchanged, and the algorithm continues processing other nodes in the open list. By continuously repeating this search and update process, the algorithm continuously expands the optimal node until the target node is successfully added to the closed list, or the open list is empty (i.e., there is no feasible path), at which point the search ends. If the target point is found, the final global planning path can be obtained by tracing back the parent node information.

[0064] As a preferred implementation, based on the A* algorithm, after obtaining the grid map information of the target environment, a dynamic heuristic function is constructed according to the obstacle density, ground slippage coefficient and moving obstacle function provided by the grid map.

[0065] In this embodiment, an improved Euclidean distance method is used to calculate the static distance based on prior information from the raster map and the travel cost, reflecting the geometric distance between the target node and the current node. The travel cost coefficient γ(n) of the raster containing the nth node in the raster map is dynamically adjusted according to the terrain type. The value of γ(n) is 1.0 for flat areas, 1.5 for slippery areas, and 2.0 for densely equipped areas. The slippery score SlipperyScore(n) ranges from 0 to 1, where 0 represents safe and 1 represents extremely dangerous. The default value of the slippery cost weight μ is 1.2.

[0066] The adaptive weight parameters α and β in the dynamic heuristic function are dynamically adjusted according to the environmental complexity. When the robot is in an open area, α is increased to prioritize static distance and shorten the search time. When the robot is in a complex area, β is increased to enhance dynamic obstacle avoidance and safety. This design allows the algorithm to automatically reduce its search speed in complex areas to avoid risks, while converging quickly in open areas, balancing efficiency and safety. The formulas for calculating α and β are as follows:

[0067] α = 1 - β.

[0068]

[0069] Where RiskScore(n) is the comprehensive risk assessment value of the nth node in the grid map, with a value of 0 to 1, k is the slope coefficient, and θ is the risk threshold with a value of 0.6.

[0070] In a preferred embodiment, step 203 can be replaced by steps 301 and 302. Wherein:

[0071] Step 301: Based on the global planning path, a path decomposition method is used to obtain multiple local targets arranged in sequence.

[0072] Step 302: For any local target, an improved dynamic window algorithm is used to obtain the local planning path from the local target to the next local target.

[0073] Furthermore, step 301 above can be replaced by steps 3011 and 3012. Wherein:

[0074] Step 3011: Obtain the sub-segments formed by each adjacent path point in the global planning path, and calculate the direction vector of each sub-segment.

[0075] Step 3012: For any sub-segment, calculate the vector angle between the direction vector of the sub-segment and the direction vector of the adjacent sub-segment. If the vector angle is greater than an angle threshold, then the common path point of the sub-segment and the adjacent sub-segment is taken as a local target.

[0076] As a preferred implementation method, such as Figure 3 As shown, based on the constructed grid map, after obtaining the global planning path using the improved A* algorithm, the turning points in the path are extracted as local targets, and then the improved DWA is used for local path planning.

[0077] In this embodiment, based on the globally planned path, the direction vectors of two adjacent sub-segments in the path are compared sequentially. If the direction changes, the common path point of these two sub-segments is taken as the turning point. All turning points are used as local targets.

[0078] An improved Direct Path Analysis (DWA) is employed, incorporating real-time environmental information for local path planning. At the starting point, the first turning point obtained from the previous steps is used as the target. The improved DWA is then used to complete local path planning to reach the first turning point. Then, the second turning point is used as the target, and the improved DWA is used to complete local path planning to reach the third turning point, and so on, until local path planning for all turning points is completed.

[0079] When the robot approaches an obstacle, the improved DWA (Path Planning and Weaving) can flexibly adjust the robot's trajectory to avoid entering the obstacle area and continue moving towards the target point. Furthermore, the improved DWA continuously updates the robot's local target and adjusts the local path accordingly to avoid collisions in real-time environmental changes and ensures optimal path adjustments based on the current obstacle situation, maintaining path safety and stability. The system determines whether the current local target has been reached. If it has, the improved DWA will determine whether the entire path planning task has been completed. If the target node has not been reached, the improved DWA continues to adjust the local path until the robot successfully reaches the target node and completes the task.

[0080] As a preferred implementation, the cross-modal fusion method is a semantic-geometric fusion method. Traditional DWA uses the same obstacle avoidance strategy for different types of obstacles, which easily gets trapped in local optima in complex environments, making it difficult to effectively avoid obstacles. To address this problem, this embodiment proposes a Semantic-Aware Dynamic Window Approach (SA-DWA), which combines the Time to Collision (TTC) method to construct the evaluation function.

[0081] The semantic information fusion module is the core of SA-DWA. By fusing data from visual inspection and depth cameras, it assigns semantic labels to obstacles. The visual inspection module provides obstacle category and boundary information through cameras, accurately identifying whether obstacles are high-risk targets (such as precision equipment, dangerous areas, etc.). The depth camera provides distance and position information of obstacles, ensuring that the robot can perceive the relative position and dynamic changes of obstacles in real time. Through the fusion of cross-modal data, a cost map is generated, on which the semantic attributes of obstacles are labeled, and the position of dynamic targets is updated in real time. The semantic labels of obstacles include not only their physical attributes (such as distance and position) but also their type information, enabling the robot to adopt differentiated obstacle avoidance strategies based on different obstacles.

[0082] In this embodiment, the obstacle avoidance strategy is further refined for static obstacles. Static obstacles are divided into different categories. Category A obstacles: infrastructure equipment. These obstacles are generally stable and do not move, so the avoidance radius for them can be relatively small. A small, conservative obstacle avoidance radius R1 = 0.5m is adopted, while a maximum permissible speed v is set. max1 =0.7m / s. Class B obstacles: precision instruments, protective devices, etc. These obstacles place slightly higher demands on the robot's avoidance capabilities and require more space. For this type of obstacle, the conservative obstacle avoidance radius is R² = 0.8m, and the maximum permissible speed is v. max2 =0.5m / s. For dynamic obstacles, because their position and velocity change constantly, the avoidance requirements are relatively more stringent. Using a larger conservative obstacle avoidance radius R3 = 1m, the maximum permissible speed is v. max3 =0.3m / s.

[0083] In this embodiment, the evaluation function of the improved dynamic window algorithm is:

[0084] cost = goal cost +clearance cost +velocity cost +σ·TTC.

[0085]

[0086] Where cost is the evaluation function, and goal is... cost The distance cost between the robot and the target point, clearance cost Velocity is the cost of maintaining a safe distance between the robot and obstacles. cost Here, x is the velocity penalty factor, TTC is the collision time between the robot and the dynamic obstacle, σ is the weighting factor corresponding to TTC, and x is the collision time between the robot and the dynamic obstacle. final Let y be the x-coordinate of the robot's current position. final Let x be the ordinate of the robot's current position. g Let y be the x-coordinate of the target point. g Let η be the ordinate of the target point, η be the weighting factor for the azimuth cost, Δθ be the angular error between the robot's current orientation and the target point, obs be a single obstacle, obstacles be a set of obstacles, max(·) be the maximum value function, d be the distance between the robot's current position and the obstacle, R be the safe avoidance radius of the obstacle, and penalty be... factor v is a penalty factor set based on the semantic labels of obstacles. candidate v is the robot's current speed. threshold This is the speed threshold.

[0087] The distance cost between the robot and the target point measures the distance between the robot's current position and the target position. The safe distance cost between the robot and obstacles considers the safe distance between the robot and the obstacle; if the distance between the robot and the obstacle is less than the set safe distance, a penalty is imposed. The speed penalty factor penalizes high speeds when the robot approaches obstacles. If the robot is close to a static obstacle and chooses a high speed, the cost function will impose a larger penalty to prevent the robot from colliding due to excessive speed.

[0088] In dynamic obstacle avoidance tasks, TTC (Total Collision Risk) is an important indicator for assessing the collision risk between the robot and obstacles. The formula for calculating TTC is as follows:

[0089]

[0090] Where TTC is the shortest time for the robot to collide with the obstacle, d l V is the relative distance between the robot and the obstacle. relative Let be the relative velocity vector between the robot and the dynamic obstacle.

[0091] In this embodiment, the speed relationship between the robot and the dynamic obstacle is as follows: Figure 4 As shown in the figure. In the figure, V1 is the robot's velocity vector, V... 1x V1 is the transverse component, V 1y V1 is the longitudinal component, V2 is the velocity vector of the dynamic obstacle, and V 2x V is the transverse component of V2, V 2y Let V2 be the longitudinal component. Through Time-to-Collapse (TTC), the robot can dynamically predict the time of collision with obstacles and take appropriate obstacle avoidance actions, such as deceleration, steering, or stopping. To calculate TTC, the component of relative velocity in the relative position direction needs to be determined. As shown in the following equation, the distance in the relative velocity direction can be obtained by calculating the dot product of the relative position vector and the relative velocity vector.

[0092] d l ·V relative =d x ·v rx +d y ·v ry .

[0093] Where, d x and d y v represents the lateral and longitudinal components of the relative distance between the robot and the obstacle, respectively. rx and v ry These are the lateral and longitudinal components of the relative velocity between the robot and the obstacle, respectively.

[0094] In path planning, calculating the Time-to-Cross (TTC) helps robots predict whether they will collide with obstacles. If the TTC is less than a preset threshold, the robot should proactively take obstacle avoidance measures, such as slowing down or stopping. If the TTC is higher, the robot can continue moving forward until it encounters a new obstacle or the environment changes. Real-time TTC calculation allows robots to dynamically adapt to environmental changes and avoid collisions, thus achieving safer and more intelligent path planning and obstacle avoidance.

[0095] In this embodiment, SA-DWA generates a dynamic window containing all possible speed combinations based on the robot's speed and acceleration constraints. Within the dynamic window, the motion trajectory for each speed combination is predicted, the cost of each trajectory is evaluated according to an evaluation function, and the trajectory with the lowest cost is selected as the current motion strategy. The TTC algorithm is used to assess the collision risk between the robot and dynamic obstacles. When the TTC is less than a certain threshold, the algorithm reduces the score on that path, thereby achieving obstacle avoidance for dynamic obstacles using TTC.

[0096] Based on the same inventive concept, this application also provides a robot path planning device for implementing the robot path planning method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more robot path planning device embodiments provided below can be found in the limitations of the robot path planning method described above, and will not be repeated here.

[0097] In another exemplary embodiment, such as Figure 5 As shown, a robot path planning device is provided, including a map building module 501, a global path acquisition module 502, a local path acquisition module 503, and a path planning module 504. The map building module 501 is used to construct a grid map of the target environment. The global path acquisition module 502 is used to obtain a globally planned path based on the grid map using an improved A* algorithm. The improved A* algorithm is based on the A* algorithm and constructs a dynamic heuristic function based on geometric distance and dynamic risk parameters. The local path acquisition module 503 is used to obtain multiple locally planned paths based on the globally planned path using a path decomposition method and an improved dynamic window algorithm. The improved dynamic window algorithm is based on the dynamic window algorithm and constructs an evaluation function based on a cross-modal fusion method and a collision time method. The path planning module 504 is used to obtain the robot path planning result based on all the locally planned paths.

[0098] In another exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 6As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs in the non-volatile storage media to run. The database stores a grid map of the target environment. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When executed by the processor, the computer program implements a robot path planning method.

[0099] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0100] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0101] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0102] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0103] The beneficial effects of this application are as follows:

[0104] To address the needs of robotic inspection in complex environments, a framework integrating global and local path planning is proposed. At the global planning level, an improved A* algorithm is employed. Through dynamic heuristic function optimization and potential field fusion strategies, combined with Euclidean distance, obstacle density, and dynamic risk parameters, an adaptive search mechanism is constructed, significantly improving the efficiency and safety of path search. Simultaneously, dynamic correction of the potential field enhances the path's adaptability to complex terrain and dense equipment. For local planning, a semantically aware dynamic window method is proposed. This method integrates visual detection and depth camera data to achieve multimodal environmental perception. Differentiated obstacle avoidance strategies are implemented for different types of obstacles, and a reinforcement learning framework is introduced to optimize the trajectory evaluation function, effectively suppressing path oscillations and improving the robustness of dynamic obstacle avoidance.

[0105] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0106] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0107] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0108] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0109] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A robot path planning method, characterized in that, The robot path planning method includes: Construct a raster map of the target environment; Based on the grid map, an improved method is adopted. The algorithm yields a globally planned path; the improvement... The algorithm in Based on the algorithm, a dynamic heuristic function is constructed using geometric distance and dynamic risk parameters; The dynamic risk parameters include obstacle density, ground slippage coefficient, and moving obstacle function; The improvements The algorithm's dynamic heuristic function is: ； ； ； ； ； ； in, For the dynamic heuristic function, The static distance is based on the raster map. This is a dynamically adjustable item. for The corresponding adaptive weight parameters, for The corresponding adaptive weight parameters, For the raster map, the first n The x-coordinate of each node For the raster map, the first n The ordinate of each node, Let x be the x-coordinate of the target node. Let be the ordinate of the target node. For the raster map, the first n The passage cost coefficient of the grid where each node is located The density of the obstacle, The number of obstacles, π Pi For statistical radius, The normalization coefficient is... The slipperiness coefficient of the ground is... Slippery rating, As a weight for slipperiness cost, For the moving obstacle function, It is a function with maximum value. To predict the trajectory and the first in the grid map n The minimum distance between nodes. As a smoothing factor, for The corresponding weighting coefficients, for The corresponding weighting coefficients, for The corresponding weighting coefficients; Based on the global planning path, multiple local planning paths are obtained by using a path decomposition method and an improved dynamic window algorithm; the improved dynamic window algorithm constructs an evaluation function based on a cross-modal fusion method and a collision time method on the basis of the dynamic window algorithm. The cross-modal fusion method is a semantic-geometric fusion method; The evaluation function of the improved dynamic window algorithm is: ； ； ； ； in, The evaluation function is... The distance cost between the robot and the target point. The cost of maintaining a safe distance between the robot and obstacles, As a speed penalty factor, TTC The collision time between the robot and the dynamic obstacle. for TTC The corresponding weighting factor, The x-coordinate of the robot's current position. Let be the ordinate of the robot's current position. Let x be the x-coordinate of the target point. Let be the ordinate of the target point. As a weighting factor for azimuth cost, The angle error between the robot's current orientation and the target point. obs For a single obstacle, obstacles A collection of obstacles. It is a function with maximum value. d The distance between the robot's current position and the obstacle. R The safe avoidance radius for obstacles. The penalty factor is set based on the semantic labels of obstacles. This represents the robot's current speed. The speed threshold; Based on all the local planning paths, the robot path planning results are obtained.

2. The robot path planning method according to claim 1, characterized in that, Based on the grid map, an improved method is adopted. After obtaining the global planned path using the algorithm, the robot path planning method further includes: The global planning path is optimized by sequentially employing the artificial potential field method and the RDP algorithm.

3. The robot path planning method according to claim 1, characterized in that, Based on the global planning path, multiple local planning paths are obtained using a path decomposition method and an improved dynamic window algorithm, specifically including: Based on the global planning path, a path decomposition method is used to obtain multiple local targets arranged in sequence; For any local target, an improved dynamic window algorithm is used to obtain the local planning path from the local target to the next local target.

4. The robot path planning method according to claim 3, characterized in that, Based on the global planning path, a path decomposition method is used to obtain multiple sequentially arranged local objectives, specifically including: Obtain the sub-segments formed by each adjacent path point in the global planning path, and calculate the direction vector of each sub-segment; For any sub-segment, calculate the vector angle between the direction vector of the sub-segment and the direction vector of the adjacent sub-segment; if the vector angle is greater than the angle threshold, then the common path point of the sub-segment and the adjacent sub-segment is taken as the local target.

5. A robot path planning device, characterized in that, The robot path planning device includes: The map building module is used to build a raster map of the target environment; The global path acquisition module is used to obtain the path based on the raster map using an improved method. The algorithm yields a globally planned path; the improvement... The algorithm in Based on the algorithm, a dynamic heuristic function is constructed using geometric distance and dynamic risk parameters; The dynamic risk parameters include obstacle density, ground slippage coefficient, and moving obstacle function; The improvements The algorithm's dynamic heuristic function is: ； ； ； ； ； ； in, For the dynamic heuristic function, The static distance is based on the raster map. This is a dynamically adjustable item. for The corresponding adaptive weight parameters, for The corresponding adaptive weight parameters, For the raster map, the first n The x-coordinate of each node For the raster map, the first n The ordinate of each node, Let x be the x-coordinate of the target node. Let be the ordinate of the target node. For the raster map, the first n The passage cost coefficient of the grid where each node is located The density of the obstacle, The number of obstacles, π Pi For statistical radius, The normalization coefficient is... The slipperiness coefficient of the ground is... Slippery rating, As a weight for slipperiness cost, For the moving obstacle function, It is a function with maximum value. To predict the trajectory and the first in the grid map n The minimum distance between nodes. As a smoothing factor, for The corresponding weighting coefficients, for The corresponding weighting coefficients, for The corresponding weighting coefficients; The local path acquisition module is used to obtain multiple local planned paths based on the global planned path, using a path decomposition method and an improved dynamic window algorithm; the improved dynamic window algorithm constructs an evaluation function based on the cross-modal fusion method and the collision time method on the basis of the dynamic window algorithm; The cross-modal fusion method is a semantic-geometric fusion method; The evaluation function of the improved dynamic window algorithm is: ； ； ； ； in, The evaluation function is... The distance cost between the robot and the target point. The cost of maintaining a safe distance between the robot and obstacles, As a speed penalty factor, TTC The collision time between the robot and the dynamic obstacle. for TTC The corresponding weighting factor, The x-coordinate of the robot's current position. Let be the ordinate of the robot's current position. Let x be the x-coordinate of the target point. Let be the ordinate of the target point. As a weighting factor for azimuth cost, The angle error between the robot's current orientation and the target point. obs For a single obstacle, obstacles A collection of obstacles. It is a function with maximum value. d The distance between the robot's current position and the obstacle. R The safe avoidance radius for obstacles. The penalty factor is set based on the semantic labels of obstacles. This represents the robot's current speed. The speed threshold; The path planning module is used to obtain the robot path planning result based on all the local planned paths.

6. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the robot path planning method according to any one of claims 1-4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the robot path planning method according to any one of claims 1-4.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the robot path planning method according to any one of claims 1-4.

Images