Automobile path planning method based on dynamic pheromone punishment coefficient and application

Abstract

The scheme provides a car path planning method based on a dynamic information element punishment coefficient and application, a navigation path is obtained through multiple iteration searches based on an optimized ant colony algorithm, a deadlock information element punishment, elite collaborative information element reward and path optimization information element reward are introduced in an information element updating formula of the ant colony algorithm, so as to improve path convergence stability and optimize path quality, and improve path planning capability of the ant colony optimization algorithm in a complex environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of vehicle path planning and autonomous navigation, in particular to a vehicle path planning method based on a dynamic pheromone penalty coefficient and application. BACKGROUND

[0002] Global path planning is to calculate the optimal macroscopic driving route from the starting point to the ending point based on high-precision map and other static prior information. There are many kinds of existing global path planning methods, such as A algorithm, Dijkstra algorithm, Rapidly-exploring Random Tree (RRT) and its optimization algorithm, etc. The A algorithm can efficiently search for the optimal path by combining heuristic functions, but its performance is heavily dependent on the design of the heuristic function; the Dijkstra algorithm can guarantee to find the globally shortest path, but it is low in computational efficiency in large-scale graph search; the RRT algorithm has probabilistic completeness in high-dimensional complex space, but the path generated by it is usually not optimal and has randomness. In summary, the above algorithms are difficult to achieve efficient calculation while ensuring path quality in large-scale, high-dimensional or unstructured road surface environment.

[0003] As an effective heuristic search method, the ant colony optimization algorithm can find an approximate optimal solution when solving path planning problems by simulating the process of ants searching for food, and has good global search ability and robustness, so it has been widely concerned. However, in the process of implementing path planning by this algorithm, affected by factors such as driving environment and algorithm defects, on the one hand, ants are prone to deadlock in the search process, i.e. trapped in "local optimum"; on the other hand, the traditional ant colony algorithm lacks effective guidance for high-quality paths, which leads to long paths.

[0004] In view of the problem that the above path planning process may be trapped in a deadlock state, patent CN116627175A proposes an improved ant colony algorithm based on a backtracking mechanism and pheromone clearing. When an ant is trapped in a deadlock point, the invention retreats one step and includes the node before the retreat into the taboo table, and clears the pheromone concentration at the deadlock point to guide the ant to escape from the deadlock area; patent number CN115560774B proposes an improved ant colony algorithm based on hierarchical punishment of backtracking steps. This method applies pheromone punishment according to the number of backtracking steps of ants to reduce the attractiveness of the deadlock path. Patent number CN202310015209.4 proposes an improved ant colony algorithm based on segment updating of deadlock paths. This method divides the path of a deadlock ant into an advantage path and a disadvantage path, and rewards and punishes the pheromones respectively, which improves the diversity of path exploration. However, the above solutions all use fixed rules to deal with deadlocks, but due to the lack of adaptive adjustment mechanism for the strength of punishment, the algorithm is prone to over-punishment or insufficient punishment in complex environments, which has poor convergence stability, weak adaptability and leads to long paths.

[0005] To solve the problem of long path caused by the lack of effective guidance of high-quality path in the traditional ant colony algorithm in the above path planning process, patent No. CN118927239A proposes an elite ant strategy based on contribution degree screening, which rewards the elite path individually; patent CN111860754B adopts a collaborative strategy of hybrid ant colony and genetic algorithm, and introduces a delta penalty value in pheromone update to reward elite ants that exceed the global optimal solution. However, although the above schemes have some innovations in elite stratification and hybrid optimization, the guidance of elite paths is still focused on part of the individuals or a single optimal solution, and the effective guidance of high-quality paths is not good.

[0006] Therefore, the current path planning method based on ant colony optimization algorithm still performs poorly in complex environments, and how to achieve efficient, safe and reliable path planning is still an important issue that unmanned vehicles need to overcome. SUMMARY

[0007] The embodiments of the present application provide a car path planning method based on dynamic pheromone penalty coefficient and application. To solve the problem of possible deadlock state and lack of effective guidance of high-quality path in the ant colony optimization algorithm in path planning, a fuzzy controller is introduced to adjust the penalty coefficient in real time, and a collaborative reward mechanism based on group consensus is introduced to improve the path convergence stability and optimize the path quality, thereby improving the path planning ability of the ant colony optimization algorithm in complex environments.

[0008] In a first aspect, the embodiments of the present application provide a path planning method for unmanned vehicles based on dynamic pheromone penalty coefficient, which includes the following steps:

[0009] S1: Construct a grid map between the navigation starting point and the navigation ending point, wherein the grid map includes feasible regions and infeasible regions;

[0010] S2: taking the navigation starting point as the starting point and the navigation ending point as the ending point, performing multi-generation iterative search based on the ant colony algorithm in the grid map to obtain a navigation path, wherein after each generation of iteration ends, for an ant that falls into a deadlock state, iteration progress and deadlock occurrence frequency of the current ant are calculated, the iteration progress and the deadlock occurrence frequency are input into a fuzzy controller to output a penalty coefficient, the penalty coefficient is substituted into a penalty formula of the current ant to obtain a deadlock pheromone penalty of edges corresponding to last two steps of the current ant; for an ant that does not fall into the deadlock state, all redundant nodes in an original path of the current ant are deleted to construct a new path, and a path optimization pheromone reward added to each edge on the new path is calculated based on the new path and the original path; an elite path set is obtained according to all paths after each generation of iteration ends, a consensus edge commonly passed through by at least two elite paths is obtained based on the elite path set, a synergistic weight of the consensus edge in the elite path set is calculated, an elite synergistic pheromone reward of the consensus edge is calculated based on the synergistic weight, and the deadlock pheromone penalty, the elite synergistic pheromone reward, and the path optimization pheromone reward are introduced into a pheromone update formula of the corresponding edge.

[0011] In a second aspect, an unmanned vehicle path planning system based on a dynamic pheromone penalty coefficient is provided, and includes:

[0012] a map construction module configured to construct a grid map between a navigation starting point and a navigation ending point, wherein the grid map includes a feasible region and an infeasible region;

[0013] a path planning module configured to take the navigation starting point as the starting point and the navigation ending point as the ending point, perform multi-generation iterative search based on the ant colony algorithm in the grid map to obtain a navigation path, wherein after each generation of iteration ends, for an ant that falls into a deadlock state, iteration progress and deadlock occurrence frequency of the current ant are calculated, the iteration progress and the deadlock occurrence frequency are input into a fuzzy controller to output a penalty coefficient, the penalty coefficient is substituted into a penalty formula of the current ant to obtain a deadlock pheromone penalty of edges corresponding to last two steps of the current ant; for an ant that does not fall into the deadlock state, all redundant nodes in an original path of the current ant are deleted to construct a new path, and a path optimization pheromone reward added to each edge on the new path is calculated based on the new path and the original path; an elite path set is obtained according to all paths after each generation of iteration ends, a consensus edge commonly passed through by at least two elite paths is obtained based on the elite path set, a synergistic weight of the consensus edge in the elite path set is calculated, an elite synergistic pheromone reward of the consensus edge is calculated based on the synergistic weight, and the deadlock pheromone penalty, the elite synergistic pheromone reward, and the path optimization pheromone reward are introduced into a pheromone update formula of the corresponding edge.

[0014] In a third aspect, the embodiments of the present application provide a readable storage medium, the readable storage medium storing a computer program, the computer program including program codes for controlling a process to execute the process, and the process includes the unmanned vehicle path planning method based on the dynamic pheromone penalty coefficient.

[0015] The main contributions and innovations of the present application are as follows:

[0016] 1. A fuzzy controller is introduced to adjust the penalty coefficient in real time, the dynamic penalty coefficient is used as the adjustment means, the penalty strength interval is divided according to the deadlock frequency and the iteration progress, the controllable penalty measures are matched, the strong punishment of early high-frequency deadlock is realized, the balance of exploration and convergence in the middle period is realized, the light punishment of occasional deadlock in the later period is realized, the solution diversity is protected, the problem that the ants are easy to fall into the deadlock state in the search process is solved, and then the precise guidance for the adjustment of the search direction of the subsequent ants is provided, and the adaptive deadlock processing is realized, the ability of the algorithm to escape from the local optimum is significantly improved, and the effect of the planned path is better.

[0017] 2. A collaborative reward mechanism based on group consensus is introduced, the common features of multiple elite paths of the same generation are focused, the frequency of each node being passed through by the elite paths is counted, the collaborative weight is calculated according to the consensus strength, and the nodes in the consensus are rewarded with additional pheromones, so as to effectively guide the high-quality paths by using the wisdom of the group, effectively avoid the accidental misdirection and the risk of local optimum of a single path, avoid the accidental misdirection of a single solution, and reduce the risk of falling into a local optimum, and the high-quality paths are effectively guided.

[0018] The details of one or more embodiments of the present application are presented in the following drawings and description to make other features, objects and advantages of the present application more apparent. BRIEF DESCRIPTION OF DRAWINGS

[0019] The drawings described herein are used to provide further understanding of the present application, and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings:

[0020] Figure 1 is a flowchart of the unmanned vehicle path planning method based on the dynamic pheromone penalty coefficient according to the embodiments of the present application;

[0021] Figure 2 is a logic diagram of the calculation of the path optimization pheromone reward and the deadlock pheromone penalty of the present scheme.

[0022] Figure 3 is a schematic diagram of adjacent grids of the present scheme.

[0023] Figure 4 is a schematic diagram of a multi-U-shaped obstacle environment.

[0024] Figure 5 is a trajectory planning path graph of a traditional ant colony algorithm.

[0025] Figure 6 is a simulation graph of the ant colony algorithm of the present solution.

[0026] Figure 7 is a hardware structure schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] The exemplary embodiments will be described in detail herein below with reference to the drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with one or more embodiments of the description. Instead, they are merely examples of apparatuses and methods consistent with some aspects of one or more embodiments of the description as detailed in the appended claims.

[0028] It should be noted that the steps of the methods in other embodiments are not necessarily performed in the order shown and described in the present description. In some other embodiments, the steps included in the methods can be more or less than those described in the present description. Furthermore, a single step described in the present description can be broken down into multiple steps in other embodiments; and multiple steps described in the present description can be combined into a single step in other embodiments.

[0029] Embodiment One

[0030] As shown in Figure 1 , the present solution provides a path planning method for an unmanned vehicle based on a dynamic pheromone penalty coefficient, comprising the following steps:

[0031] S1: Construct a grid map between a navigation starting point and a navigation ending point, wherein the grid map includes a feasible region and an infeasible region;

[0032] S2: taking the navigation starting point as the starting point and the navigation ending point as the ending point, performing multi-generation iterative search based on the ant colony algorithm in the grid map to obtain a navigation path, wherein after each generation of iteration ends, for an ant that falls into a deadlock state, the iteration progress and the deadlock occurrence frequency of the current ant are calculated, the iteration progress and the deadlock occurrence frequency are input into a fuzzy controller to output a penalty coefficient, the penalty coefficient is substituted into a penalty formula of the current ant to obtain a deadlock pheromone penalty of edges corresponding to the last two steps of the current ant; for an ant that does not fall into a deadlock state, all redundant nodes in the original path of the current ant are deleted to construct a new path, and based on the new path and the original path, a path optimization pheromone reward is calculated for each edge added to the new path; an elite path set is obtained according to all paths after each generation of iteration ends, and a consensus edge commonly passed through by at least two elite paths is obtained based on the elite path set, a synergistic weight of the consensus edge in the elite path set is calculated, an elite synergistic pheromone reward of the consensus edge is calculated based on the synergistic weight, and the deadlock pheromone penalty, the elite synergistic pheromone reward, and the path optimization pheromone reward are introduced into a pheromone update formula of the corresponding edge.

[0033] In step S1, the feasible region of the grid map refers to a region in which the vehicle can travel, and the infeasible region refers to a region in which the vehicle cannot travel. The feasible region and the infeasible region are divided so as to facilitate subsequent path planning by the ant colony algorithm.

[0034] In some embodiments, the navigation map is intercepted based on the navigation starting point and the navigation ending point, and the navigation map is rasterized to obtain a plurality of grids to obtain a grid map.

[0035] It should be noted that the rasterization processing refers to dividing the navigation map according to a*a grid size.

[0036] Further, the roads in the navigation map are extracted, and the grid attributes of each grid are defined based on the relative positional relationship between the roads and each grid. If the current grid is in the road, the grid attribute corresponding to the current grid is feasible, and all grids with the feasible grid attribute constitute the feasible region. If the grid is not in the road, the grid attribute corresponding to the current grid is infeasible, and all grids with the infeasible grid attribute constitute the infeasible region.

[0037] It should be noted that the grid attribute corresponding to the current grid is defined as infeasible when the grid is partially in the road. That is, only when the grid is completely in the road, the grid attribute corresponding to the current grid is feasible.

[0038] In other words, the S1 of the present scheme intercepts the range required for navigation in the high-precision map as a navigation map based on the navigation starting point and the navigation ending point, extracts the road from the navigation map and performs rasterization processing on the navigation map to obtain a raster map, and defines all the grids as being in the road or not in the road according to the road, wherein the grids in the road constitute a feasible area that allows the vehicle to travel, and the grids not in the road constitute an infeasible area that does not allow the vehicle to travel.

[0039] In step S2, the present scheme optimizes the pheromone update formula on the basis of the traditional ant colony algorithm, introduces the deadlock pheromone punishment, the elite cooperative pheromone reward and the path optimization pheromone reward into the pheromone update formula, avoids the accumulation of pheromones of the deadlock path, and strengthens the high-quality direction through double forward rewards, so as to quickly converge to the effective path and improve the exploration ability of the better path.

[0040] In step S2, the ant is used as a path search agent, each grid in the raster map is used as a node, the number of ants, the maximum number of iterations, the pheromone importance factor and the heuristic information importance factor are set, and the ant is placed in the node where the navigation starting point is located and starts multi-generation iteration search based on the state transition probability formula, wherein the pheromone concentration in the state transition probability formula is updated according to the pheromone update formula.

[0041] In other words, each ant is used as an independent path search agent to simulate the foraging behavior of natural ants, and the grids with feasible attributes in the raster map are used as the nodes of path search, the initialization of the basic parameters of the ant colony search is completed, and then the multi-generation iteration search is started. After each generation of iteration is completed, all ants complete the construction of the path from the navigation starting point to the navigation ending point, and update the pheromone concentration according to the pheromone update formula for the next generation of iteration.

[0042] In some embodiments, the path with the maximum number of iterations is obtained as the navigation path.

[0043] The state transition probability formula is as follows: ;

[0044] wherein i, j are nodes, allowed represents that the grid attribute of the node is feasible, is the pheromone concentration from the i node to the j node at time t, , is the heuristic function from the i node to the j node at time t, α is the pheromone importance factor, and β is the heuristic information importance factor.

[0045] wherein the heuristic function is: ;

[0046] wherein is the distance between the i node and the j node.

[0047] wherein the pheromone update formula is:

[0048]

[0049] wherein is a preset pheromone evaporation factor, is the pheromone concentration from the i node to the j node at time t, is the global pheromone concentration from the i node to the j node at time t+1, is a pheromone evaporation process, is the total sum of pheromone added by all ants from the i node to the j node, wherein k represents an ant, and m is the number of ants, is the elite collaborative pheromone reward from the i node to the j node, is the path optimization pheromone reward from the i node to the j node, is the deadlock pheromone penalty from the i node to the j node.

[0050] It should be noted that the pheromone update formula is for updating the pheromone between the i node and the j node, and if there is no elite collaborative pheromone reward, path optimization pheromone reward, or deadlock pheromone penalty between the i node and the j node, the corresponding value is 0.

[0051] Figure 2 is a logic diagram for calculating the path optimization pheromone reward and the deadlock pheromone penalty of the present scheme.

[0052] Regarding the deadlock pheromone penalty:

[0053] After each iteration of the ant colony algorithm, all ants need to be judged for the deadlock state, and the deadlock pheromone penalty is calculated for the ants that fall into the deadlock state.

[0054] Specifically, the adjacent grids of the grid where the ant is currently located are checked, and if all the adjacent grids meet the preset conditions, the current ant falls into the deadlock state, wherein the preset conditions are any of exceeding the grid map boundary, the grid attribute being unfeasible, or having been visited by the current ant.

[0055] It should be noted that the adjacent grids refer to all grids adjacent to the current grid, i.e., the possible moving nodes of the ant.

[0056] In some embodiments, the adjacent grids are the up, down, left, right, and four diagonal grids of the current grid.

[0057] As Figure 3 ​As shown, the present scheme performs triple verification on each adjacent grid: (1) whether it exceeds the grid map boundary; (2) whether the grid attribute is infeasible; and (3) whether it has been accessed by the current ant. If all adjacent grids at least do not satisfy one of the above conditions, it indicates that the feasible field of the current grid is empty, corresponding to the current ant being in a deadlock state.

[0058] Further, for the ant trapped in the deadlock state, the iteration progress and the deadlock occurrence frequency of the current ant are calculated, the iteration progress and the deadlock occurrence frequency are input into the fuzzy controller to output a penalty coefficient, and the penalty coefficient is substituted into the penalty formula of the current ant to obtain the deadlock pheromone penalty of the edge corresponding to the last two steps of the current ant.

[0059] In some embodiments, the iteration progress is the ratio of the current iteration number to the maximum iteration number, and the deadlock occurrence frequency is the ratio of the number of ants in the deadlock state in the current iteration to the total number of ants.

[0060] In some embodiments, the penalty coefficient rule is defined in the fuzzy controller, and the iteration progress and the deadlock occurrence frequency are input into the fuzzy controller to match the corresponding penalty coefficient rule to obtain the penalty coefficient.

[0061] Specifically, the penalty coefficient rule is: when the deadlock occurrence frequency is greater than 0.3 and the iteration progress is less than 0.4, the penalty coefficient is any value in 0.8-1; when the deadlock occurrence frequency is not greater than 0.3 and not less than 0.1, and the iteration progress is not less than 0.4 and not greater than 0.7, the penalty coefficient is any value in 0.4-0.7; when the deadlock occurrence frequency is less than 0.1 and the iteration progress is greater than 0.7, the penalty coefficient is any value in 0.1-0.3.

[0062] The penalty coefficient rule table of the corresponding fuzzy controller is shown in Table 1 as follows:

[0063]

[0064] Further, the penalty coefficient is substituted into the penalty formula of the current ant to obtain the deadlock pheromone penalty of the last two steps of the current ant, and the formula is as follows:

[0065] where Q is a preset pheromone enhancement coefficient, is the path length of the ant in the deadlock state, is the penalty coefficient, is the deadlock pheromone penalty of the edge of the last two steps of the current ant, wherein the last two steps correspond to the e node to the a node.

[0066] It should be noted that the path length of the ant in the deadlock state is obtained by directly obtaining the number of rows of the path matrix of the ant (i.e. the number of nodes in the path).

[0067] Regarding the path optimization pheromone reward:

[0068] The present solution simplifies the path of the ant not trapped in the deadlock state, that is, it judges whether the path of the ant not trapped in the deadlock state is redundant, and if there is a redundant node, all redundant nodes in the original path of the current ant are deleted to construct a new path, and the path optimization pheromone reward is calculated based on the new path and the original path.

[0069] Specifically, for the path of each ant not trapped in the deadlock state, three nodes on the original path are sequentially traversed in node order from the starting node, and for each three nodes, the first node, the second node and the third node are defined according to the node order, and if there is a grid with a feasible grid attribute on the connecting line between the first node and the third node, the second node is defined as a redundant node, and all redundant nodes in the original path of the current ant are deleted to construct a new path.

[0070] For example, three nodes on the original path are taken from the starting node, which are a0, a1, and a2, and a0 and a2 are connected. If there is no obstacle on the connecting line, a1 is a redundant node, which is deleted and the node number is updated, and the node a2 after the change becomes the node a1, and the detection of a1, a2, and a3 for redundant nodes continues, and then all nodes are sequentially traversed until all redundant nodes are deleted to obtain a new path.

[0071] Further, in the step of calculating the path optimization pheromone reward based on the new path and the original path, the path improvement rate of the new path relative to the original path is calculated, and if the path improvement rate is greater than the improvement rate threshold, the path optimization pheromone reward is calculated based on the path improvement rate.

[0072] In some embodiments, the path improvement rate is as follows:

[0073] ;

[0074] wherein is the length of the original path, is the length of the new path.

[0075] It should be noted that the path improvement rate ranges from 0 to 1, and the greater the path improvement rate, the better the optimization effect.

[0076] In some embodiments, the path optimization pheromone reward is as follows:

[0077] wherein The pheromone reward for the edge (p, h) to be added to the new path, is a preset reward coefficient, Q is a preset pheromone enhancement coefficient, is the length of the new path.

[0078] The elite collaborative pheromone reward is:

[0079] The scheme proposes a collaborative reward mechanism based on group consensus, which selects the elite path of each generation, identifies the consensus edge shared by the elite path, and rewards the collaborative pheromone according to the consensus strength.

[0080] In some embodiments, all paths are sorted by path length from short to long, and the first multiple paths are obtained as the elite path set.

[0081] In some embodiments, the first 20% of the paths are obtained as the elite path set.

[0082] It should be noted that the path at this time is the path after the redundancy processing.

[0083] In some embodiments, the number of times each edge is passed by the elite path in the elite path set is calculated, and if the number of times is greater than 2, the current edge is defined as a consensus edge, and the collaborative weight of each consensus edge in the elite path set is calculated, and the calculation formula is as follows:

[0084]

[0085] wherein is the number of times of the consensus edge (c, w), and m is the number of elite paths in the elite path set, is the collaborative weight, The greater the value, the more the current consensus edge is recognized by more elite paths.

[0086] In some embodiments, the formula for calculating the elite collaborative pheromone reward is as follows:

[0087] wherein is a preset collaborative reward coefficient, Q is a preset pheromone constant, is the average length of all elite paths in the elite path set, is the elite collaborative pheromone reward for the node c to the node w added to the consensus.

[0088] In order to verify the effectiveness of the present application, the traditional ant colony algorithm and the ant colony algorithm of the present scheme are simulated in a multi-U-shaped obstacle environment by MATLAB software, and the multi-U-shaped obstacle environment is as shown in Figure 4 , the trajectory planning path diagram of the traditional ant colony algorithm obtained is as shown in Figure 5 , and the simulation diagram of the ant colony algorithm of the present scheme is as shown inFigure 6 As shown.

[0089] From the simulation results, it can be seen that the path planned by the traditional ant colony algorithm often contains a large number of redundant nodes, so that the path is redundant and not smooth, and the path planned by the scheme is obviously shortened in length compared with the path planned by the traditional ant colony algorithm.

[0090] Embodiment two

[0091] Based on embodiment one, the scheme provides an unmanned vehicle path planning system based on a dynamic pheromone penalty coefficient, comprising:

[0092] A map construction module is configured to construct a grid map between a navigation starting point and a navigation ending point, wherein the grid map includes a feasible region and an infeasible region.

[0093] A path planning module is configured to take the navigation starting point as a starting point and the navigation ending point as an ending point, and perform a multi-generation iterative search based on an ant colony algorithm in the grid map to obtain a navigation path. After each iteration ends, for an ant that falls into a deadlock state, the iteration progress and the deadlock occurrence frequency of the current ant are calculated, the iteration progress and the deadlock occurrence frequency are input into a fuzzy controller to output a penalty coefficient, the penalty coefficient is substituted into a penalty formula of the current ant to obtain a deadlock pheromone penalty of edges corresponding to the last two steps of the current ant. For an ant that does not fall into a deadlock state, all redundant nodes in the original path of the current ant are deleted to construct a new path, and a path optimization pheromone reward is calculated based on the new path and the original path and added to each edge of the new path. According to all paths after each iteration ends, an elite path set is obtained, and a consensus edge commonly passed through by at least two elite paths is obtained based on the elite path set. The synergistic weight of the consensus edge in the elite path set is calculated, the elite synergistic pheromone reward of the consensus edge is calculated based on the synergistic weight, and the deadlock pheromone penalty, the elite synergistic pheromone reward, and the path optimization pheromone reward are introduced into a pheromone update formula of the corresponding edge.

[0094] The content of this embodiment two is the same as that of embodiment one and is not repeated here.

[0095] Embodiment three

[0096] The embodiment also provides an electronic device, which refers to Figure 7 including a memory 404 and a processor 402, the memory 404 stores a computer program, and the processor 402 is configured to run the computer program to perform the steps in any one of the above-mentioned embodiments of the method for planning a path of an unmanned vehicle based on a dynamic pheromone penalty coefficient.

[0097] Specifically, the processor 402 can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement one or more embodiments of the present application. The memory 404 can include a mass storage for data or instructions. The memory 404 can be used to store or cache various data files needed for processing and / or communication, and possible computer program instructions executed by the processor 402.

[0098] The processor 402 implements any one of the above-described unmanned vehicle path planning methods based on dynamic pheromone penalty coefficients by reading and executing computer program instructions stored in the memory 404.

[0099] Optionally, the electronic device can further include a transmission device 406 connected to the processor 402, and an input / output device 408 connected to the processor 402.

[0100] The input / output device 408 is used to input or output information. In the present embodiment, the input information can be a navigation starting point and a navigation ending point, etc., and the output information can be a navigation path, etc.

[0101] Optionally, in the present embodiment, the processor 402 can be configured to perform the following steps by computer program:

[0102] S1: constructing a grid map between the navigation starting point and the navigation ending point, wherein the grid map includes feasible regions and infeasible regions;

[0103] S2: taking the navigation starting point as the starting point and the navigation ending point as the ending point, performing multi-generation iterative search based on the ant colony algorithm in the grid map to obtain a navigation path, wherein after each generation iteration ends, for an ant trapped in a deadlock state, an iteration progress of the current ant and a deadlock occurrence frequency are calculated, the iteration progress and the deadlock occurrence frequency are input into a fuzzy controller to output a penalty coefficient, the penalty coefficient is substituted into a penalty formula of the current ant to obtain a deadlock pheromone penalty of edges corresponding to last two steps of the current ant; for an ant not trapped in the deadlock state, all redundant nodes in an original path of the current ant are deleted to construct a new path, and based on the new path and the original path, a path optimization pheromone reward is calculated for each edge added to the new path; an elite path set is obtained according to all paths after each generation iteration ends, and at least two consensus edges commonly passed through by the elite paths are obtained based on the elite path set, a synergistic weight of the consensus edges in the elite path set is calculated, an elite synergistic pheromone reward of the consensus edges is calculated based on the synergistic weight, and the deadlock pheromone penalty, the elite synergistic pheromone reward and the path optimization pheromone reward are introduced into a pheromone update formula of the corresponding edges

[0104] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation manners, and this embodiment will not be repeated here.

[0105] Generally, various embodiments can be implemented in hardware or special-purpose circuitry, software, logic or any combination thereof. Some aspects of the application can be implemented in hardware, while other aspects can be implemented using firmware or software in conjunction with a controller, microprocessor or other computing device, but the application is not limited thereto. While various aspects of the application can be illustrated and described as block diagrams, flow charts, or using some other pictorial representation, it is well understood that these blocks, apparatus, systems, techniques or methods described herein can be implemented in hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.

[0106] Embodiments of the application can be implemented by computer software executable by a data processor of the mobile device such as in the processor entity, or by hardware, or by a combination of software and hardware. Computer software or program, also called program product, including software routines, applets and / or macros, can be stored in any apparatus-readable data storage medium and they include program instructions to implement certain tasks. The program product can include one or more computer-executable components tangibly embodied in a computer- readable medium, when executed, for implementing one or more embodiments of the present application. The one or more computer-executable components can be one or more of: a process; a function; a routine; a sag; a subroutine; a plug-in; an app; or a program. The one or more computer-executable components can include a computer program that is written in any form of programming language, including code.

[0107] Those skilled in the art should clearly understand that each technical feature in the above embodiments can be combined arbitrarily, and for the sake of brevity, each technical feature in the above embodiments is not described in all possible combinations, however, as long as the combination of technical features does not exist, it should be considered as the scope of the present application.

[0108] The above embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be understood as a limitation on the scope of the present application. It should be pointed out that for ordinary skilled in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A path planning method for autonomous vehicles based on dynamic pheromone penalty coefficient, characterized in that, Includes the following steps: S1: Construct a grid map between the navigation start point and the navigation end point, where the grid map includes feasible and infeasible areas; S2: Starting from the navigation start point and ending at the navigation end point, a multi-generation iterative search based on the ant colony algorithm is performed on the grid map to obtain the navigation path. After each iteration, for ants trapped in deadlock, the iteration progress and deadlock frequency of the current ant are calculated. The iteration progress and deadlock frequency are input into the fuzzy controller to output the penalty coefficient. The penalty coefficient is substituted into the penalty formula of the current ant to obtain the deadlock pheromone penalty for the edge corresponding to the last two steps of the current ant. For ants not trapped in deadlock, all redundant nodes in the original path of the current ant are deleted to construct a new path. The path optimization pheromone reward is calculated for each edge on the new path based on the new path and the original path. An elite path set is obtained based on all paths after each iteration. A consensus edge that is traversed by at least two elite paths is obtained based on the elite path set. The collaborative weight of the consensus edge in the elite path set is calculated. The elite collaborative pheromone reward of the consensus edge is calculated based on the collaborative weight. The deadlock pheromone penalty, elite collaborative pheromone reward, and path optimization pheromone reward are introduced into the pheromone update formula of the corresponding edge.

2. The autonomous vehicle path planning method based on dynamic pheromone penalty coefficient according to claim 1, characterized in that, Check the adjacent grids of the grid where the ant is currently located. If all adjacent grids meet the preset conditions, the current ant is in a deadlock state. The preset conditions are that the ant exceeds the grid map boundary, the grid attribute is infeasible, or the ant has already visited any of the grids.

3. The autonomous vehicle path planning method based on dynamic pheromone penalty coefficient according to claim 1, characterized in that, The fuzzy controller defines penalty coefficient rules. The iteration progress and deadlock frequency are input into the fuzzy controller to match the corresponding penalty coefficient rules and obtain the penalty coefficient.

4. The autonomous vehicle path planning method based on dynamic pheromone penalty coefficient according to claim 1, characterized in that, The deadlock pheromone penalty for the last two steps of the current ant is: ； Where Q is the preset pheromone enhancement coefficient. Let be the path length of the ant that is deadlocked. The penalty coefficient is... The deadlock pheromone penalty for the last two steps of the current ant's edge.

5. The autonomous vehicle path planning method based on dynamic pheromone penalty coefficient according to claim 1, characterized in that, For each ant's path that is not deadlocked, traverse the original path sequentially from the starting node and take three nodes. For each set of three nodes, define them as the first node, the second node, and the third node according to the node order. If there is a grid with a feasible grid attribute on the connection between the first node and the third node, then the second node is defined as a redundant node. Delete all redundant nodes in the original path of the current ant and construct a new path.

6. The autonomous vehicle path planning method based on dynamic pheromone penalty coefficient according to claim 1, characterized in that, Calculate the path improvement rate of the new path relative to the original path. If the path improvement rate is greater than the improvement rate threshold, calculate the path optimization pheromone reward based on the path improvement rate.

7. The autonomous vehicle path planning method based on dynamic pheromone penalty coefficient according to claim 1, characterized in that, The pheromone reward for path optimization is as follows: ; ； in The length of the original path. R is the length of the new path, and R is the path improvement rate. in To optimize pheromone rewards for edges that will be added to the new path, Q is the preset reward coefficient, and Q is the preset pheromone enhancement coefficient. This represents the length of the new path.

8. The autonomous vehicle path planning method based on dynamic pheromone penalty coefficient according to claim 1, characterized in that, The formula for elite synergistic pheromone reward is as follows: ; in Let m be the number of consensus edges (c, w), and m be the number of elite paths in the elite path set. For collaborative weights, Here, Q is a preset collaborative reward coefficient, and Q is a preset pheromone constant. The average length of all elite paths in the elite path set. Elite collaborative pheromone rewards are given to node c to node w when added to the consensus edge.

9. A path planning system for autonomous vehicles based on a dynamic pheromone penalty coefficient, characterized in that, include: The map building module is used to build a raster map located between the navigation start point and the navigation end point, where the raster map includes feasible and infeasible areas; The path planning module uses the navigation starting point as the starting point and the navigation ending point as the ending point to perform multi-generation iterative search based on the ant colony algorithm in a grid map to obtain the navigation path. After each iteration, for ants trapped in a deadlock state, the iteration progress and deadlock frequency of the current ant are calculated. The iteration progress and deadlock frequency are input into the fuzzy controller to output the penalty coefficient. The penalty coefficient is substituted into the penalty formula of the current ant to obtain the deadlock pheromone penalty for the edge corresponding to the last two steps of the current ant. For ants not trapped in a deadlock state, all redundant nodes in the original path of the current ant are deleted to construct a new path. The path optimization pheromone reward is calculated for each edge on the new path based on the new path and the original path. An elite path set is obtained from all paths after each iteration. Based on the elite path set, consensus edges traversed by at least two elite paths are obtained. The collaborative weight of the consensus edge in the elite path set is calculated. The elite collaborative pheromone reward of the consensus edge is calculated based on the collaborative weight. The deadlock pheromone penalty, elite collaborative pheromone reward, and path optimization pheromone reward are introduced into the pheromone update formula of the corresponding edge.

10. A readable storage medium, characterized in that, The readable storage medium stores a computer program, the computer program including program code for controlling a process to execute the process, the process including the autonomous vehicle path planning method based on dynamic pheromone penalty coefficient according to any one of claims 1 to 8.

Images