Intelligent driving logistics vehicle path planning method and device without map navigation

Abstract

The invention discloses an intelligent driving logistics vehicle path planning method and device without map navigation, and belongs to the technical field of path planning, and the method comprises the steps: obtaining an initialized grid map of a vehicle; searching in the initialized grid map according to a heuristic search algorithm to determine a predicted path of the vehicle based on the starting point and the ending point; acquiring a plurality of historical prediction paths of the vehicle and a plurality of corresponding historical positions of the vehicle; the relative position relations between the historical positions of the vehicles and the current position of the vehicles are obtained respectively, and position alignment is conducted on the historical prediction paths and the prediction paths according to the relative position relations; performing weighted calculation on the plurality of historical prediction paths after position alignment and the prediction path to obtain a target path of the vehicle; according to the method, the plurality of historical predicted paths and the predicted paths of the vehicle are subjected to weighted calculation, and the motion states of the vehicle in the continuous time period and the corresponding path planning results are compared, so that the path of the vehicle is corrected in time, and the problem of sudden change of the path is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and in particular to a method and device for path planning of an intelligent driving logistics vehicle without map navigation. Background Art

[0002] With the development of e-commerce and intelligent manufacturing, the demand for automated logistics vehicles in logistics warehouses and distribution centers is increasing day by day. Traditional AGVs usually rely on a preset map for path planning, which requires the environment to be relatively static and known. However, in practical applications, the logistics environment often changes dynamically, including changes in the positions of goods, the emergence of temporary obstacles, and adjustments to work processes, etc. Therefore, traditional map-based methods are difficult to cope with these uncertainties, resulting in low efficiency or even inability to work properly.

[0003] However, most existing mapless path planning schemes ignore the influence of the time dimension and fail to fully consider the motion state of the logistics vehicle within a continuous time period and its influence on future path selection, resulting in the problem that the driving path of the vehicle changes suddenly due to sudden changes in the external environment during the driving process.

[0004] Therefore, in the process of path planning for vehicles without a map in the prior art, there is a problem that the path changes suddenly due to the failure to consider the time change. Summary of the Invention

[0005] In view of this, it is necessary to provide a method and device for path planning of an intelligent driving logistics vehicle without map navigation to solve the problem that in the process of path planning for vehicles without a map in the prior art, the path changes suddenly due to the failure to consider the time change.

[0006] To solve the above problems, in a first aspect, the present invention provides a method for path planning of an intelligent driving logistics vehicle without map navigation, including: Obtain the initial grid map of the vehicle; Search and determine the predicted path of the vehicle based on the starting point and the ending point in the initial grid map according to the heuristic search algorithm; Obtain multiple historical predicted paths of the vehicle and the corresponding multiple historical positions of the vehicle; Respectively obtain the relative position relationships between the multiple historical positions of the vehicle and the current position of the vehicle, and align the positions of the multiple historical predicted paths and the predicted path according to the relative position relationships; Perform weighted calculation on the multiple historical predicted paths and the predicted path after position alignment to obtain the target path of the vehicle.

[0007] In some possible implementation manners, respectively obtain the relative position relationships between multiple vehicle historical positions and the current vehicle position, and perform position alignment on the multiple historical prediction paths and the prediction path according to the relative position relationships, including: Determine the relative coordinate differences between multiple vehicle historical positions and the current vehicle position based on a pre-constructed coordinate system; Use the current vehicle position as the origin, and respectively determine the coordinates of each point on the prediction path based on the relative position relationship between the current vehicle position and the prediction path; Respectively use the relative coordinate differences as the coordinate point positions of multiple vehicle historical positions, and respectively determine the coordinates of each point on the multiple historical prediction paths based on the relative position relationships between the multiple vehicle historical positions and the multiple historical prediction paths.

[0008] In some possible implementation manners, perform weighted calculation on the multiple historical prediction paths and the prediction path after position alignment to obtain the target path of the vehicle, including: Respectively determine the path grid weights of the grids corresponding to each path according to the path weights of the multiple historical prediction paths and the prediction path; Obtain the paths corresponding to each grid in the initialized grid map after position alignment, and calculate the score matrix of the grids by weighted calculation according to the path grid weights; Based on the starting point and the ending point, select the continuous grid points with the highest score matrix values as the target path.

[0009] In some possible implementation manners, obtain the paths corresponding to each grid in the initialized grid map after position alignment, and calculate the score matrix of the grids by weighted calculation according to the path grid weights, including: Based on the timestamps of multiple vehicle historical positions and the timestamp of the current vehicle position, respectively calculate multiple initial weights of the multiple historical prediction paths according to the weight calculation formula; Perform normalization processing on the multiple initial weights to obtain multiple path weights of the multiple historical prediction paths, and respectively determine the path grid weights of the grids based on each path according to the path weights; Based on the path grid weights, respectively calculate the score matrix of each grid in the initialized grid map after alignment according to the score matrix calculation formula.

[0010] In some possible implementation manners, the weight calculation formula is:

[0011] Wherein, is the initial weight of the i th historical prediction path, is the decay rate parameter, is the iThe difference between the timestamps of the historical positions and the timestamp of the current position of the vehicle; The calculation formula for normalization is:

[0012] Wherein, is the path grid weight of the grid corresponding to the i th historical predicted path; The calculation formula for the score matrix of each grid is:

[0013] Wherein, is the score matrix of the position , is the th path, and is the existence status of the position

[0014] in the path. If the position exists in the path, it is 1; otherwise, it is 0. Obtaining multiple vehicle historical positions corresponding to multiple timestamps with the smallest time difference from the timestamp of the current position of the vehicle; Determining that the prediction path results corresponding to multiple vehicle historical positions are multiple historical prediction paths.

[0015] In some possible implementation manners, determining that the prediction path results corresponding to multiple vehicle historical positions are multiple historical prediction paths includes: Obtaining a preset number of multiple vehicle historical positions and multiple historical prediction paths, and storing the multiple vehicle historical positions and the multiple historical prediction paths in a cache area; For each current position of the vehicle, it is required to obtain a preset number of multiple vehicle historical positions, and when the current position of the vehicle changes, the cache area is updated correspondingly.

[0016] In some possible implementation manners, when the current position of the vehicle changes, correspondingly updating the cache area further includes: Determining the coordinate position of the changed vehicle position according to the rotation angle between the changed vehicle position and the current position of the vehicle, and deleting the vehicle historical position and the historical prediction path corresponding to the timestamp with the largest time difference from the timestamp of the current position of the vehicle in the cache area; Wherein, the coordinate conversion formula for determining the coordinate position of the changed vehicle position according to the rotation angle is:

[0017] Tran( ) represents the coordinate conversion function,(x a ，y a ) is the representation of a certain point p in the A coordinate system corresponding to the current position of the vehicle. (x b ，y b ) is the representation of point p in the B coordinate system corresponding to the changed position of the vehicle. (positive on the left and negative on the right) is the rotation angle from the A coordinate system to the B coordinate system. (x b0 ，y b0 ) is the representation of the origin of the B coordinate system in the A coordinate system.

[0018] In some possible implementation manners, searching and determining a predicted path of the vehicle based on a starting point and an ending point in an initialized grid map according to a heuristic search algorithm includes: Constructing an open list and a closed list, the closed list includes the starting point, and the open list includes all other points of the initialized grid map; Based on the total cost calculation formula and the ending point, determining the adjacent node corresponding to the minimum total cost based on the starting point in the open list as the first target point, and transferring the first target point from the open list to the closed list; Judging whether the closed list contains the ending point; If so, determining the predicted path based on the closed list; If not, based on the total cost calculation formula and the ending point, determining the adjacent node corresponding to the minimum total cost based on the first target point in the open list as the second target point, and transferring the second target point from the open list to the closed list, and repeating the iteration until the closed list contains the ending point.

[0019] In a second aspect, the present invention further provides a path planning device for an intelligent driving logistics vehicle without a map, including: A grid map initialization module, configured to obtain an initialized grid map of the vehicle; A path prediction module, configured to search and determine a predicted path of the vehicle based on a starting point and an ending point in the initialized grid map according to a heuristic search algorithm; A historical data acquisition module, configured to acquire multiple historical predicted paths of the vehicle and corresponding multiple vehicle historical positions; A position alignment module, configured to respectively obtain the relative position relationships between multiple vehicle historical positions and the current position of the vehicle, and perform position alignment on the multiple historical predicted paths and the predicted path according to the relative position relationships; A path planning module is used to perform weighted calculations on multiple historical prediction paths and prediction paths after position alignment to obtain the target path of the vehicle.

[0020] The beneficial effects of adopting the above embodiments are as follows: An intelligent driving logistics vehicle path planning method without a map provided by the present invention realizes real-time dynamic update of the vehicle's external environment information by obtaining the initial grid map of the vehicle, so as to timely and accurately obtain reliable prediction paths; By performing weighted calculations on multiple historical prediction paths and prediction paths of the vehicle, comparing the motion states of the vehicle in consecutive time periods and the corresponding path planning results, the path of the vehicle is corrected in a timely manner, avoiding the problem of path mutation. Description of the Drawings

[0021] Figure 1 It is a flowchart of an embodiment of the intelligent driving logistics vehicle path planning method without a map provided by the present invention; Figure 2 It is a flowchart of an embodiment of training the high-reflectivity removal network model provided by the present invention; Figure 3 It is a flowchart of an embodiment of calculating the comprehensive loss value provided by the present invention; Figure 4 It is a flowchart of an embodiment of obtaining multi-scale features of a feature map at different scales provided by the present invention; Figure 5 It is a structural diagram of an embodiment of the multi-head attention mechanism provided by the present invention; Figure 6 It is a structural block diagram of an embodiment of the intelligent driving logistics vehicle path planning device without a map provided by the present invention. Detailed Embodiments

[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0023] It should be understood that the schematic drawings are not drawn to scale. The flowcharts used in the present invention illustrate operations implemented according to some embodiments of the present invention. It should be understood that the operations in the flowchart may not be implemented in sequence, and steps without logical context relationships may be reversed or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart or remove one or more operations from the flowchart under the guidance of the content of the present invention. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor systems and / or microcontroller systems.

[0024] In the embodiments of the present invention, the descriptions such as "first" and "second" involved are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Therefore, the technical features defined with "first" and "second" may explicitly or implicitly include at least one of such features.

[0025] Reference to "embodiment" herein means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present invention. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0026] In order to solve the problem that in the process of path planning for a vehicle without a map in the prior art, there is a path mutation due to the failure to consider time changes, the present invention provides a path planning method and device for an intelligent driving logistics vehicle without a map, which will be described in detail below.

[0027] As Figure 1 shown, Figure 1 is a schematic flowchart of an embodiment of the path planning method for an intelligent driving logistics vehicle without a map provided by the present invention, including: S101: Obtain the initial grid map of the vehicle; In some embodiments of the present invention, the grid map is a grid-based map that uses discrete cells to represent the world. The value of each cell represents the information of the corresponding area in the real world. The resolution of the map depends on the size of the cells. The smaller the cells, the higher the accuracy of the map.

[0028] In the initialized grid map, with the current position of the vehicle as the origin, the position where the vehicle is located is represented by data, realizing the adaptive annotation of the environment where the vehicle is located without precise positioning information.

[0029] S102: Search and determine the predicted path of the vehicle based on the starting point and the ending point in the initialized grid map according to the heuristic search algorithm; In some embodiments of the present invention, the heuristic search algorithm includes the A* algorithm, the greedy algorithm, the ant colony algorithm, the genetic algorithm, the simulated annealing algorithm, etc. These algorithms use heuristic information to guide the search to reduce the search scope and improve the efficiency.

[0030] The A* algorithm (A-star Search Algorithm) is a heuristic search algorithm that finds the shortest path from a given starting point to one or more ending points in a graph search problem. A* combines the characteristics of Dijkstra and greedy best-first search, and estimates the total cost of reaching the target node from a node n through the total cost calculation formula, f(n) = g(n) + h(n) where, f(n) is the total cost of node n , g(n) is the actual cost from the starting point to node n , h(n) is the heuristic function, which is the estimated cost from node n to the ending point.

[0031] The heuristic function h(n) should accurately reflect the real distance from node n to the ending point as much as possible, but cannot overestimate, otherwise the guarantee of the optimal solution may be lost. Therefore, the Euclidean distance is used as the heuristic function.

[0032] Through the A* algorithm, the estimated cost of any point can be calculated, and then the path planning can be guided to obtain the predicted path of the vehicle based on any point.

[0033] S103: Obtain multiple historical predicted paths of the vehicle and the corresponding multiple historical positions of the vehicle; In some embodiments of the present invention, the starting point of each historical predicted path is the historical position of the vehicle, that is, with the historical position of the vehicle at each moment as the starting point, the predicted path of the vehicle at that moment is determined.

[0034] S104: Respectively obtain the relative position relationships between multiple historical positions of the vehicle and the current position of the vehicle, and align the positions of multiple historical predicted paths and the predicted path according to the relative position relationships; S105: Perform weighted calculation on the multiple historical predicted paths and the predicted path after position alignment to obtain the target path of the vehicle.

[0035] In this embodiment, by obtaining the initial grid map of the vehicle, the external environment information of the vehicle is updated in real time and dynamically, so as to obtain a reliable prediction path in a timely and accurate manner; by performing weighted calculations on multiple historical prediction paths and prediction paths of the vehicle, comparing the motion states of the vehicle within consecutive time periods and the corresponding path planning results, the path of the vehicle is corrected in a timely manner, avoiding the problem of path mutation.

[0036] In some embodiments of the present invention, in S101, in order to obtain the initial grid map of the vehicle, it is necessary to obtain it in combination with the sensors installed on the vehicle itself, which will not be elaborated here.

[0037] It should be noted that the initial grid map includes both the current position information of the vehicle, the destination information of the vehicle, and the obstacle information in the environment where the vehicle is located, that is, the information data of the positions that the vehicle cannot reach, so as to meet the subsequent requirements for path planning of the vehicle.

[0038] In some embodiments of the present invention, in S102, in order to search and determine the prediction path of the vehicle based on the starting point and the ending point in the initial grid map according to the heuristic search algorithm, as Figure 2 shown, Figure 2 is a schematic flowchart of an embodiment for determining the prediction path provided by the present invention, including: S201: Construct an open list and a closed list. The closed list includes the starting point, and the open list includes all other points in the initial grid map; In a specific embodiment, an open list (Open Set) and a closed list (Closed Set) are used to manage and track the nodes in the search process, helping the algorithm effectively determine which node to explore next and avoiding repeated processing of nodes that have been checked. The open list contains all discovered but unprocessed nodes. The closed list contains all processed nodes that have been fully evaluated, that is, the neighboring nodes of these nodes have been checked and no longer need to be visited again.

[0039] S202: Based on the total cost calculation formula and the ending point, determine the neighboring node corresponding to the minimum total cost based on the starting point in the open list as the first target point, and transfer the first target point from the open list to the closed list; In a specific embodiment, calculate the total cost values of all nodes in the open list, put the starting node A into the closed list (Closed Set), mark the node corresponding to the minimum total cost in the open list as the first target point A1, pop the A1 node from the open list, and add it to the closed list.

[0040] Further, search for neighborhood nodes that can be reached from node A1 and are not in the closed list. For each neighborhood node, the specific formula for calculating the new cost of reaching neighborhood node H through A1 is as follows: g(A 1 -H)=g(A 1 )+cost(A 1 ，H) Wherein, g(A 1 -H) is the new cost of reaching neighborhood node H through A1, g(A 1 ) is the actual cost of moving from the starting point to A1, cost(A 1 ，H) is the estimated cost of moving from A1 to H.

[0041] S203: Determine whether the closed list contains the end point; S204: If so, determine the predicted path based on the closed list; S205: If not, determine the neighborhood node corresponding to the minimum total cost based on the first target point in the open list as the second target point according to the total cost calculation formula and the end point, and transfer the second target point from the open list to the closed list, and repeat the iteration until the closed list contains the end point.

[0042] In a specific embodiment, during the repeated iteration process, the starting node A is added to the open list, and its adjacent grid cells H are checked. The parent node of each neighborhood node is the starting node A. If the surrounding grid cell is marked as an obstacle or it is already in the closed list, it is skipped; otherwise, it is added to the open list.

[0043] In a specific embodiment, the predicted path of the vehicle is obtained by connecting all the nodes in the closed list adjacent to each other according to their positional relationships.

[0044] In this embodiment, the total cost of each node is calculated in real time according to the open list and the closed list, realizing real-time update of the cost data of the nodes, being able to respond to environmental changes in a timely manner, and since the cost of each node is the lowest, the cost of the finally formed predicted path is also the lowest, realizing real-time acquisition of the optimal predicted path of the vehicle through the A* algorithm.

[0045] In some embodiments of the present invention, in S103, in order to obtain multiple historical predicted paths of the vehicle and the corresponding multiple vehicle historical positions, first, obtain multiple vehicle historical positions corresponding to multiple time stamps with the smallest time difference from the time stamp of the current position of the vehicle; then, determine the predicted path results based on the multiple vehicle historical positions as multiple historical predicted paths.

[0046] In this embodiment, obtaining the vehicle's historical positions based on the vehicle's current position with the timestamp as the reference can ensure the continuity between historical positions, and thus ensure that the final multiple historical prediction paths are based on the latest at the current moment, improving the reliability of the historical prediction paths.

[0047] In some embodiments of the present invention, in S104, in order to respectively obtain the relative position relationships between multiple vehicle historical positions and the vehicle's current position, and align the positions of multiple historical prediction paths and the prediction path according to the relative position relationships, as Figure 3 shown, Figure 3 is a schematic flowchart of an embodiment for coordinate alignment of multiple historical prediction paths and a prediction path provided by the present invention, including: S301: Determine the relative coordinate differences between multiple vehicle historical positions and the vehicle's current position based on a pre-constructed coordinate system; S302: Use the vehicle's current position as the origin, and respectively determine the coordinates of each point of the prediction path based on the relative position relationship between the vehicle's current position and the prediction path; S303: Respectively use the relative coordinate differences as the coordinate point positions of multiple vehicle historical positions, and respectively determine the coordinates of each point of multiple historical prediction paths based on the relative position relationships between multiple vehicle historical positions and multiple historical prediction paths.

[0048] In this embodiment, by constructing a coordinate system to determine the normalized data comparison standard, the historical position coordinates of multiple vehicle historical positions in the current coordinate system are respectively determined according to the relative coordinate differences between multiple vehicle historical positions and the vehicle's current position. That is to say, the coordinate values of each historical position relative to the current coordinate system are determined; then, the coordinates of each point of multiple historical prediction paths are respectively determined based on the relative position relationships between multiple vehicle historical positions and multiple historical prediction paths, realizing filling all the coordinate data of historical prediction paths into the coordinate system.

[0049] In some embodiments of the present invention, in S105, in order to perform weighted calculation on multiple historical prediction paths and the prediction path after position alignment to obtain the target path of the vehicle, as Figure 4 shown, Figure 4 is a schematic flowchart of an embodiment for obtaining the target path of the vehicle provided by the present invention, including: S401: Respectively determine the path grid weights of the grids corresponding to each path according to the path weights of multiple historical prediction paths and the prediction path; S402: Obtain the paths corresponding to each grid in the initialized grid map after position alignment, and calculate the score matrix of the grids by weighted calculation according to the path grid weights; S403: Based on the starting point and the ending point, select the continuous grid points with the highest score matrix value as the target path.

[0050] In some embodiments of the present invention, in S401, in order to calculate the score matrix of the grid by weighted calculation according to the path grid weight, as Figure 5 shown, Figure 5 is a schematic flowchart of an embodiment for determining the score matrix of each grid provided by the present invention, including: S501: Based on the timestamps of multiple vehicle historical positions and the timestamp of the vehicle's current position, calculate multiple initial weights of multiple historical prediction paths respectively according to the weight calculation formula; In some embodiments of the present invention, the weight calculation formula is:

[0051] wherein, is the initial weight of the i th historical prediction path, is the decay rate parameter, is the difference between the timestamp of the i th historical position and the timestamp of the vehicle's current position.

[0052] It should be noted that, can be adjusted according to the actual situation. A larger will cause the weight of the earlier path to drop rapidly, while a smaller will make the influence of the earlier path last longer.

[0053] S502: Normalize the multiple initial weights to obtain multiple path weights of multiple historical prediction paths, and determine the path grid weight of each grid based on each path according to the path weight; In some embodiments of the present invention, the specific formula for normalizing the multiple initial weights is:

[0054] wherein, is the path grid weight of the grid corresponding to the i th historical path, N is the total number of historical prediction paths.

[0055] S503: Based on the path grid weight, calculate the score matrix of each grid in the aligned initialized grid map respectively according to the score matrix calculation formula.

[0056] In some embodiments of the present invention, when applying linear weighted calculation to fuse paths, it is necessary to ensure that the path planning results of the current frame and the historical frame are in the same coordinate system. For each grid coordinate (i, j) , the weighted calculation can be performed according to its appearance frequency and the corresponding weight in all paths. Then for a given set of pathsP={p 1 ，p 2 ，...，p N } , the score matrix of each grid The calculation formula is as follows:

[0057] wherein, is the score matrix at position , is the existence status of the position in the th path. If the position exists in the path, it is 1; otherwise, it is 0.

[0058] It should be noted that the same grid in this application may correspond to different path grid weights. When multiple paths pass through a certain grid, the grid corresponds to multiple path grid weights. By superimposing multiple path grid weights, the score matrix of the grid is obtained.

[0059] In this embodiment, by setting the initial weight of the historical prediction path, the importance of the historical prediction path can be adaptively corrected, so as to determine the nodes of the final target path according to actual needs, and improve the reliability of the target path.

[0060] In some embodiments of the present invention, in order to determine that the prediction path results based on multiple vehicle historical positions are multiple historical prediction paths. First, obtain a preset number of multiple vehicle historical positions and multiple historical prediction paths, and store the multiple vehicle historical positions and multiple historical prediction paths in a cache area; then, for each vehicle current position, it is required to obtain a preset number of multiple vehicle historical positions, and when the vehicle current position changes, update the cache area correspondingly.

[0061] In this embodiment, only storing a preset number of vehicle historical positions and historical prediction paths in the cache area can reduce the demand for the storage capacity of the cache area while meeting the data requirements.

[0062] Furthermore, in the process of correspondingly updating the cache area when the vehicle current position changes, it is also necessary to determine the coordinate position of the changed vehicle position according to the rotation angle between the changed vehicle position and the vehicle current position, and delete the vehicle historical position and historical prediction path corresponding to the time stamp with the largest time difference from the time stamp of the vehicle current position in the cache area; wherein, the coordinate conversion formula for determining the coordinate position of the changed vehicle position according to the rotation angle is:

[0063] Tran( ) represents the coordinate conversion function(x a ，y a ) is the representation of a certain point p in the A coordinate system corresponding to the current position of the vehicle. (x b ，y b ) is the representation of point p in the B coordinate system corresponding to the changed vehicle position. (positive on the left and negative on the right) is the rotation angle from the A coordinate system to the B coordinate system. (x b0 ，y b0 ) is the representation of the origin of the B coordinate system in the A coordinate system.

[0064] In this embodiment, based on the rotation angle generated during the vehicle change process, coordinate transformation is performed on the changed vehicle position and its corresponding predicted path, so as to ensure that all paths are based on the same coordinate system in the coordinate system and avoid the problem of data disorder.

[0065] In a specific embodiment, specifically: Create a historical data cache structure for storing past path planning results; Create a cache structure Q containing N - 1 cache units, where each cache unit is represented as q t , and each cache unit includes a time stamp T, a starting point target (i 0 ，j 0 ) , target point coordinates (i g ，j g ) , front wheel angle δ t , vehicle speed v t and path planning data PD t .

[0066] The time stamp T is used to record the time when the path planning occurs. The path data PD t consists of a series of grid coordinates and represents the path from the starting point to the target point, where t is the corresponding time stamp.

[0067] Furthermore, construct a new cache unit q t , and the new cache unit q tAdd it to the cache structure Q; The judgment condition is |Q| < N - 1. After using the A* algorithm to generate the preliminary path planning result, it is judged whether the judgment condition is satisfied. If it is satisfied, directly use the path planning result generated by the A* algorithm to construct a new cache unit q t , if it is not satisfied, fuse the preliminary path planning result with the historical planning result in the cache structure, and create the fused path planning result as a new cache unit q t , t is the time stamp at the current moment, fill in the time stamp T, the starting point target (i 0 ，j 0 ) , the coordinates of the target point (i g ，j g ) , the turning radius of the vehicle R t , the front wheel steering angle δ t , the vehicle speed v t and the path planning data PD t .

[0068] When the path planning at time t is completed and a new cache unit is created q t , judge the size of the current cache structure queue |Q| and the maximum cache quantity N - 1. If |Q| < N - 1, directly add the new cache unit q t to the end of the queue. If the cache is full, that is, |Q| = N - 1, it is necessary to remove the earliest added entry q t-9 , and then add the new cache unit q t to the head of the queue.

[0069] As a further technical solution, the method further includes: Convert the historical frame data stored in the cache to the coordinate system of the current frame; Coordinate transformation between the historical frame and the current frame:

[0070] wherein, (positive on the left and negative on the right) is the rotation angle from coordinate system A to coordinate system B. (x b0 ，y b0 ) is the representation of the origin of coordinate system B in coordinate system A. (xa ，y a ) is the representation of a certain point p in the A coordinate system, (x b ，y b ) is the representation of point p in the B coordinate system.

[0071] Then,

[0072] Without knowing the global coordinate system and without considering the non - linear dynamics of the vehicle, the coordinate transformation relationship can be approximately derived:

[0073] Where, represents the change in the front - wheel steering angle within the time interval between two frames, L is the wheelbase of the vehicle, K s is the steering stiffness, sign () is the sign function used to determine the direction of the front - wheel steering angle, Δx represents the change in displacement of the vehicle in the forward direction of the vehicle, Δy represents the change in displacement of the vehicle in the direction perpendicular to the forward direction, (x t-1 ，y t-1 ) 、 (x t ，y t ) are respectively the global coordinates of the vehicle at t - 1 and t moments, δ t-1 , δ t is the front - wheel steering angle, v t-1 , v t is the vehicle speed.

[0074] Extract data of nine frames from the cache structure Q Q(q t-9 ，q t-8 ，……，q t-1 ） and convert the path - planning data except the current frame to the current - frame coordinate system.

[0075] Using a linear weighted calculation method, the data of the current frame and historical frames are fused to obtain a score matrix S(i, j） , finally, from the score matrix S(i, j) a new path is extracted, and the sequence of consecutive grid points with the highest score can be selected as the final path to obtain the final path planning result q t , and it is updated to the cache structure Q.

[0076] In this embodiment, by obtaining the initial grid map of the vehicle, the external environment information of the vehicle is updated in real time and dynamically, so as to obtain a reliable prediction path in a timely and accurate manner; by performing weighted calculations on multiple historical prediction paths and prediction paths of the vehicle, comparing the motion states of the vehicle in consecutive time periods and the corresponding path planning results, the path of the vehicle is corrected in a timely manner, avoiding the problem of path mutation. By setting the initial weight of the historical prediction path, the importance of the historical prediction path can be adaptively corrected, so as to determine the nodes of the final target path according to actual needs and improve the reliability of the target path.

[0077] To better implement the path planning method for a mapless navigation intelligent driving logistics vehicle in the embodiments of the present invention, correspondingly, the embodiments of the present invention also provide a path planning device for a mapless navigation intelligent driving logistics vehicle, as Figure 6 shown Figure 6 is a structural block diagram of an embodiment of the path planning device for a mapless navigation intelligent driving logistics vehicle provided by the present invention. The path planning device 600 for a mapless navigation intelligent driving logistics vehicle includes: A grid map initialization module 601, configured to obtain the initial grid map of the vehicle; A path prediction module 602, configured to search and determine the prediction path of the vehicle based on the starting point and the ending point in the initial grid map according to the heuristic search algorithm; A historical data acquisition module 603, configured to acquire multiple historical prediction paths of the vehicle and the corresponding multiple vehicle historical positions; A position alignment module 604, configured to respectively obtain the relative position relationships between multiple vehicle historical positions and the current position of the vehicle, and perform position alignment on multiple historical prediction paths and prediction paths according to the relative position relationships; A path planning module 605, configured to perform weighted calculations on the multiple historical prediction paths and prediction paths after position alignment to obtain the target path of the vehicle.

[0078] The path planning device 600 for the mapless navigation intelligent driving logistics vehicle provided by the above embodiments can implement the technical solutions described in the embodiments of the path planning method for the mapless navigation intelligent driving logistics vehicle. For the specific implementation principles of the above modules or units, reference can be made to the corresponding content in the embodiments of the path planning method for the mapless navigation intelligent driving logistics vehicle, which will not be elaborated here.

[0079] The path planning method and device for the mapless navigation intelligent driving logistics vehicle provided by the present invention have been introduced in detail above. Specific examples are used in this article to elaborate on the principles and implementation manners of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A path planning method for an intelligent driving logistics vehicle without map navigation, characterized in that, Including: Obtain the initial grid map of the vehicle; Search and determine the predicted path of the vehicle based on the starting point and the ending point in the initial grid map according to the heuristic search algorithm; Obtain multiple historical predicted paths of the vehicle and corresponding multiple vehicle historical positions; Respectively obtain the relative position relationships between the multiple vehicle historical positions and the current position of the vehicle, and align the positions of the multiple historical predicted paths and the predicted path according to the relative position relationships; Perform weighted calculation on the multiple historical predicted paths and the predicted path after position alignment to obtain the target path of the vehicle.

2. The path planning method for a mapless navigation intelligent driving logistics vehicle according to claim 1, characterized in that The step of respectively obtaining the relative position relationships between the multiple vehicle historical positions and the current position of the vehicle, and aligning the positions of the multiple historical predicted paths and the predicted path according to the relative position relationships includes: Determine the relative coordinate differences between the multiple vehicle historical positions and the current position of the vehicle based on a pre-constructed coordinate system; Take the current position of the vehicle as the origin, and respectively determine the coordinates of each point of the predicted path based on the relative position relationship between the current position of the vehicle and the predicted path; Respectively take the relative coordinate differences as the coordinate point positions of the multiple vehicle historical positions, and respectively determine the coordinates of each point of the multiple historical predicted paths based on the relative position relationships between the multiple vehicle historical positions and the multiple historical predicted paths.

3. The path planning method of the mapless navigation intelligent driving logistics vehicle according to claim 2, characterized in that, The step of performing weighted calculation on the multiple historical predicted paths and the predicted path after position alignment to obtain the target path of the vehicle includes: Respectively determine the path grid weights of the grids corresponding to each path according to the path weights of the multiple historical predicted paths and the predicted path; Obtain the paths corresponding to each grid in the initial grid map after position alignment, and calculate the score matrix of the grid by weighted calculation according to the path grid weights; Based on the starting point and the ending point, select the continuous grid points with the highest score matrix value as the target path.

4. The path planning method for a mapless navigation intelligent driving logistics vehicle according to claim 3, wherein The step of obtaining the paths corresponding to each grid in the initial grid map after position alignment, and calculating the score matrix of the grid by weighted calculation according to the path grid weights includes: Based on the timestamps of the multiple vehicle historical positions and the timestamp of the current position of the vehicle, respectively calculate the multiple initial weights of the multiple historical predicted paths according to the weight calculation formula; Perform normalization processing on the multiple initial weights to obtain the multiple path weights of the multiple historical predicted paths, and respectively determine the path grid weights of the grid based on each path according to the path weights; Based on the path grid weights, respectively calculate the score matrix of each grid in the aligned initial grid map according to the score matrix calculation formula.

5. The path planning method of the mapless navigation intelligent driving logistics vehicle according to claim 4, characterized in that, The weight calculation formula is: Among them, is the i initial weight of the th historical prediction path, is the attenuation rate parameter, i is the difference between the timestamp of the th historical position and the timestamp of the current position of the vehicle; The calculation formula for performing normalization processing is: Among them, is the i path grid weight of the grid corresponding to the th historical prediction path. The calculation formula for the score matrix of each grid is: Among them, is the score matrix of the position . is the existence status of the position in the th path. If the position exists in the path, it is 1; otherwise, it is 0.

6. The path planning method of the mapless navigation intelligent driving logistics vehicle according to claim 1, characterized in that The step of obtaining the multiple historical predicted paths of the vehicle and corresponding multiple vehicle historical positions includes: Obtain the multiple vehicle historical positions corresponding to multiple timestamps with the smallest time differences from the timestamp of the current position of the vehicle; Determine that the prediction path results corresponding to the multiple vehicle historical positions are the multiple historical prediction paths.

7. The path planning method of the mapless navigation intelligent driving logistics vehicle according to claim 6, characterized in that The determination that the prediction path results corresponding to the multiple vehicle historical positions are the multiple historical prediction paths includes: Obtain a preset number of the multiple vehicle historical positions and the multiple historical prediction paths, and store the multiple vehicle historical positions and the multiple historical prediction paths in a cache area; For each current vehicle position, only a preset number of the multiple vehicle historical positions need to be obtained, and when the current vehicle position changes, the cache area is updated correspondingly.

8. The path planning method of the mapless navigation intelligent driving logistics vehicle according to claim 7, characterized in that, The corresponding update of the cache area when the current vehicle position changes further includes: Determine the coordinate position of the changed vehicle position according to the rotation angle between the changed vehicle position and the current vehicle position, and delete the vehicle historical position and the historical prediction path corresponding to the time stamp with the largest time difference from the time stamp of the current vehicle position in the cache area; Wherein, the coordinate transformation formula for determining the coordinate position of the changed vehicle position according to the rotation angle is: Tran( ) represents a coordinate transformation function, (x a ，y a ) is the representation of a certain point p in the A coordinate system corresponding to the current position of the vehicle, (x b ，y b ) is the representation of point p in the B coordinate system corresponding to the changed position of the vehicle, (positive on the left and negative on the right) is the rotation angle from the A coordinate system to the B coordinate system, (x b0 ，y b0 ) is the representation of the origin of the B coordinate system in the A coordinate system.

9. The path planning method for a mapless navigation intelligent driving logistics vehicle according to claim 1, wherein The search and determination of the prediction path of the vehicle based on the starting point and the ending point in the initialized grid map according to the heuristic search algorithm includes: Construct an open list and a closed list, the closed list includes the starting point, and the open list includes all other points in the initialized grid map; Based on the total cost calculation formula and the ending point, determine that the adjacent node corresponding to the minimum total cost based on the starting point in the open list is the first target point, and transfer the first target point from the open list to the closed list; Judge whether the closed list contains the ending point; If so, determine the prediction path based on the closed list; If not, based on the total cost calculation formula and the ending point, determine that the adjacent node corresponding to the minimum total cost based on the first target point in the open list is the second target point, and transfer the second target point from the open list to the closed list, and repeat the iteration until the closed list contains the ending point.

10. An intelligent driving logistics vehicle path planning device for mapless navigation, characterized in that, Includes: A grid map initialization module, used to obtain the initialized grid map of the vehicle; A path prediction module, used to search and determine the prediction path of the vehicle based on the starting point and the ending point in the initialized grid map according to the heuristic search algorithm; A historical data acquisition module, used to obtain the multiple historical prediction paths of the vehicle and the corresponding multiple vehicle historical positions; A position alignment module, used to respectively obtain the relative position relationships between the multiple vehicle historical positions and the current vehicle position, and perform position alignment on the multiple historical prediction paths and the prediction path according to the relative position relationships; A path planning module, used to perform weighted calculation on the multiple historical prediction paths and the prediction path after position alignment to obtain the target path of the vehicle.

Images