A multi-rovers collaborative mission planning method and system for lunar exploration time network

Abstract

The present invention discloses a multi-rover collaborative task planning method and system for a lunar surface detection time network, and relates to the field of lunar surface detection technology. The technical points of the present invention include: modeling based on the lunar surface multi-rover collaborative detection task; establishing a passable road network between lunar surface detectable points based on the lunar surface collaborative detection task; establishing a multi-rover lunar surface detection time network based on the detection information of the lunar surface passable road network; using a parthenogenetic algorithm to solve the path of the lunar surface collaborative detection task, and obtaining the optimal path for each rover to perform lunar surface collaborative detection. The present invention effectively solves the problem of large-scale lunar surface multi-rover task planning, and the detection task volume of each rover can be well distributed, which significantly reduces the overall detection time.

Description

Technical Field

[0001] The present invention relates to the technical field of lunar exploration, and in particular to a multi-rover collaborative task planning method and system for a lunar exploration time network. Background Art

[0002] As human exploration of the moon continues to deepen, future lunar patrol missions will develop in the direction of larger scope and higher precision. The lunar rover of the US Apollo 15 was mainly driven by astronauts, with the longest single driving distance of 12.47 kilometers and the farthest distance from the lunar module of 5 kilometers. The Yutu-2 lunar rover in China's Chang'e-4 lunar exploration mission has ground remote control and autonomous navigation capabilities, but its range of movement is small and its speed is slow. The United States is currently conducting research on a new generation of lunar patrol robots, including the Intrepid lunar rover program, the All-Terrain Hex-Limbed Extra-Terrestrial Explorer program, and the Space Exploration Vehicle, etc., aiming to develop patrol robots with larger ranges and higher autonomous capabilities.

[0003] With the development of my country's lunar exploration project, the main tasks of lunar exploration in the future will revolve around the establishment of unmanned or manned lunar scientific research stations on the moon. The site selection of the lunar laboratory is one of the most important preliminary preparations for the construction of the station. It is necessary to examine the topography, geology, lunar soil carrying capacity, mineral resources and water ice enrichment, light, electromagnetic radiation, thermal environment and other multiple constraints of the station address. It is necessary to carefully survey many alternative sites for selection. The distance between the alternative addresses may reach hundreds of kilometers and there are many of them. Therefore, the detection mission of the alternative addresses puts forward higher requirements for the large-scale coordinated detection of multiple patrols. Facing such future scenarios that emphasize coordinated patrols, reasonable and effective patrol detection task allocation is of great significance to improving task efficiency. Summary of the invention

[0004] In view of the above problems, for the large-scale, multi-task, multi-rover collaborative exploration mission on the lunar surface, the present invention proposes a multi-rover collaborative task planning method and system for the lunar exploration time network, so as to improve the connectivity of the lunar road network, so that the roving vehicles have more optional paths, thereby reducing the overall exploration time to a certain extent.

[0005] According to one aspect of the present invention, a multi-rovers collaborative mission planning method for a lunar exploration time network is proposed, the method comprising:

[0006] S1. Modeling based on the collaborative exploration mission of multiple lunar rovers;

[0007] S2. Based on the lunar surface collaborative exploration mission, establish a passable road network between the lunar surface detectable points;

[0008] S3. Establish a multi-rovers lunar surface detection time network based on the detection information of the lunar surface traversable road network;

[0009] S4. Use the partheno-genetic algorithm to solve the path of the lunar collaborative exploration mission and obtain the optimal path for each rover to conduct collaborative exploration of the lunar surface.

[0010] Furthermore, the specific steps of S1 include:

[0011] Suppose there are m rovers on the moon, one of which sets out at a fixed time T, and needs to detect n mission points in total. Each mission point only needs to be detected once. The detection time of the rover at the mission point is fixed, and the rover needs to be charged regularly during the journey; each rover lands directly at a different mission point and starts the detection mission with the landing point as the starting point. After the detection is completed, it does not need to return to the starting point, and the total detection time is required to be the shortest. The multi-traveling salesman model of the above problem is described on a weighted undirected graph G = (V, E), let the set V = {v1, v2, …, v n} represents the vertex set of graph G, and the set E={ei j =(v i ,v j ):v i ,v j ∈V,i≠j} represents the set of edges of graph G, using b1, b2, …, b m represents m patrollers, A={a1,a2,…,a n –1, a n} is a set of task points; define the following variables:

[0012]

[0013]

[0014] Formula (1) indicates that when patroller k moves from task point v i Departure, arrive at the mission point v j When , the variable takes the value of 1, otherwise it takes the value of 0; Formula (2) means that when patroller k has visited task point i, the variable takes the value of 1, otherwise it takes the value of 0;

[0015] Using variable d ij Represents the task point v i to v j The distance between ij It is defined as the length of the path between two task points calculated by the path search algorithm; t ij Represents the time from the task point v i Depart to the mission point vj Total time to end, t ij Including the patroller from the task point v i To the mission point v j The time spent on the journey and at the mission point v j The time required to complete the detection is used as the time required to connect the nodes v in the topology graph. i With node v j The edge ij The length of the edge in the weighted undirected graph G is calculated as:

[0016] e ij =t ij =t ij (d ij ,v,t C ,t E ) (3)

[0017] t ij The distance d between two points ij , patrol speed v, detection time t C , Charging time t E The objective function of multi-rovers collaborative detection mission planning is:

[0018] t=min(max(t 1 ,t 2 ,...,t m )) (4)

[0019] in:

[0020]

[0021] The constraints are:

[0022]

[0023] Formula (4) indicates that the total time taken by m patrollers to complete the task is the minimum; Formula (5) indicates the time required for the patroller launched for the kth time to complete the task; Formula (6) indicates that starting from the virtual starting point v0, all detection points are visited by at least one patroller once.

[0024] Furthermore, the specific steps of S2 include:

[0025] First, the lunar elevation map is converted into a binary map of the feasible area. The establishment of the binary map of the feasible area requires extracting terrain information from the lunar DEM map data and calculating the slope. Then, the map positions with slopes exceeding the rover's capacity limit are regarded as obstacles, and the positions with slopes less than the rover's maximum capacity limit are regarded as feasible areas.

[0026] Next, a drivable road map between detection points is established based on the binary map of the feasible area, including determining detection points and calculating drivable roads.

[0027] Furthermore, the specific steps of determining the detection points and calculating the passable roads in S2 include:

[0028] For the location of the detection point, the requirements of the detection mission and the lunar terrain information are taken into consideration, and the detection point is selected outside the obstacle-dense area while meeting the detection mission;

[0029] For the problem of calculating the feasibility between detection points, the A* algorithm is used to search for paths between detection points on the binary map of the feasible area. If there is a path between two detection points, it is considered that the two nodes are passable. During the search process, the node with the highest priority is selected from the candidate node list as the next traversal node each time; Manhattan distance is used as the heuristic function in the A* algorithm to estimate the path cost between the current node and the target node. Manhattan distance is the sum of the distance between the two nodes in the y-axis direction and the distance in the x-axis direction.

[0030] Furthermore, the specific step of determining the detection point in S2 also includes:

[0031] The maximum single moving radius r of the patroller is limited according to its driving capability. That is, when the straight-line distance between two detection points is greater than r, it is considered that the two points are not directly reachable. If a detection point cannot reach any node within the radius r, it is considered that the detection point is unreachable and is removed from the road network.

[0032] Furthermore, the specific steps of S3 include:

[0033] First, for terrain information, the density of obstacles around the path is used to evaluate the difficulty of the path. A rectangular sliding window is used to translate on the path to obtain the density of obstacles at each point on the path.

[0034] The relationship between the patrol vehicle's speed and obstacle density is expressed as:

[0035]

[0036] Where b h The obstacle density threshold represents the minimum movement speed of the patroller. When the obstacle density is greater than the threshold, the patroller moves at a constant speed at the minimum speed. l The obstacle density threshold represents the maximum movement speed of the patroller. When the obstacle density is less than the threshold, the patroller moves at a constant speed at the maximum speed.

[0037] After obtaining the patroller's speed, calculate the time required for the patroller to complete a path without stopping. The calculation formula is:

[0038]

[0039] In the formula, represents the time required for the patroller to complete the path from detection point i to detection point j without stopping; l represents the patroller movement distance corresponding to a grid in the map; P ij Represents the set of map grid points corresponding to the path from detection point i to detection point j. When node i and node j are not adjacent, the path between the nodes is searched on the established lunar road network by Dijkstra algorithm;

[0040] The calculation formula for the patroller's charging time in a path is expressed as:

[0041]

[0042] In the formula, represents the time taken by the patroller to charge on the path from detection point i to detection point j; t c Indicates the time required to fully charge; d ij represents the path length from detection point i to detection point j, d ij =l×size(P ij ), size(P ij ) refers to the path P ij The number of grid points; d c Indicates the distance the patroller travels between two charges;

[0043] When the patroller arrives at the detection point, it is assumed that the patroller immediately starts the detection operation, and the operation time at each detection point is set to Assuming that the detector is fully charged after detection and before moving to the next detection point, and the charging time is included in the detection time; the total time taken by the detector from detection point i to detection point j is expressed as:

[0044]

[0045] Furthermore, the calculation formula for obstacle density in S3 is:

[0046]

[0047] In the formula, x n ,y n Represents the x, y coordinates of the current path point n; m, n represent the width and height of the sliding window; M(x, y) represents the value at the coordinates of the point (x, y) on the map.

[0048] Further, the specific steps of S4 include:

[0049] Encoding: A two - segment chromosome encoding method is adopted for the expression of solutions. The n lunar exploration nodes and m rovers are encoded into a chromosome consisting of two parts. The first segment is a chromosome of length n - 1, where the n lunar exploration nodes are represented by n real numbers 1, 2, …, n, representing that the rover starts from the same starting point and there are n - 1 remaining exploration nodes to traverse. The second segment is m - 1 positive integers. The n lunar exploration nodes are divided into m segments, respectively representing the lunar exploration nodes to be traversed by m rovers. The values are the positions where the chromosome is broken. The value of the breakpoint determines the number of nodes that each rover needs to detect. The numbers in the second segment are arranged in ascending order from small to large, and the difference between adjacent numbers is not less than the minimum number of nodes detected by the rover.

[0050] Population initialization: There are n lunar exploration nodes, so the lunar exploration nodes are numbered from 1 to n. Excluding the starting node 1, individuals with randomly generated chromosome genes randomly arranged from 2 to n and n - 1 breakpoints are generated. Here, the arrangement is the visiting order of each rover.

[0051] Fitness function: The collaborative detection objective function is adopted as the fitness function of the genetic algorithm; when the total time taken for all rovers to complete all lunar explorations is the least, the fitness of this path is the highest.

[0052] Selection operation: Randomly select h individuals from the current population, calculate their fitness according to the above - mentioned fitness function, and store the individual with the highest fitness in a temporary population containing h individuals, and the other h - 1 individuals are removed. The individual with the highest fitness replaces them through the mutation operation.

[0053] Mutation operation: Randomly generate 2 mutation positions I and J, where I < J, and the insertion position P of the mutation segment, and use different methods to mutate each individual in the aforementioned temporary population.

[0054] Iteration termination: When the set termination condition is met, stop the iterative search. At this time, the solution is the optimal path for each rover to conduct collaborative detection.

[0055] Further, the different methods of using different methods to mutate each individual in the aforementioned temporary population include: not performing any operation, gene inversion, gene transposition, gene left - shift, gene right - shift, and only modifying the breakpoint of the latter part without modifying the former part of the individual; among them, resetting the breakpoint can be used in combination with gene inversion, gene transposition, and gene shift respectively.

[0056] According to another aspect of the present invention, a multi-rover collaborative mission planning system for a lunar exploration time network is provided, the system comprising:

[0057] A model building module configured to model a multi-rovers collaborative exploration mission on the lunar surface;

[0058] A traversable road network establishment module, which is configured to establish a traversable road network between traversable points on the lunar surface based on the lunar surface collaborative exploration mission;

[0059] A detection time network establishment module, which is configured to establish a multi-rover lunar surface detection time network based on the detection information of the lunar surface traversable road network;

[0060] The optimal path calculation module is configured to perform path planning for the lunar collaborative exploration mission based on a partheno-genetic algorithm to obtain the optimal path for each rover to perform lunar collaborative exploration.

[0061] The beneficial technical effects of the present invention are:

[0062] 1) Aiming at the large-scale collaborative detection mission of multiple rovers on the lunar surface, the present invention establishes a multi-traveling salesman problem model that meets the constraints of the collaborative detection mission, and constructs the lunar road network and the lunar detection time network respectively. Finally, the partheno-genetic algorithm is used to solve the multi-rovers collaborative detection problem, which can effectively solve the large-scale multi-rovers mission planning problem on the lunar surface, and the detection task volume of each rover can be better allocated.

[0063] 2) As the departure time interval increases, the maximum detection time of multiple patrollers increases approximately linearly, and patrollers that depart later will be automatically assigned fewer detection tasks. The planning algorithm can balance the time of each patroller so that multiple patrollers can complete their tasks at a similar time, thereby efficiently allocating the task volume of each patroller.

[0064] 3) When the number of detection nodes remains the same, increasing the number of patrollers can reduce the workload of each patroller, thereby significantly reducing the overall detection time.

[0065] 4) Increasing the maximum single movement radius of the rover can improve the connectivity of the lunar road network, providing the rover with more optional paths, thereby reducing the overall detection time to a certain extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The present invention can be better understood by referring to the description given below in conjunction with the accompanying drawings, which together with the following detailed description are included in this specification and form a part of this specification, and are used to further illustrate the preferred embodiments of the present invention and explain the principles and advantages of the present invention.

[0067] Figure 1The present invention is a flowchart of a method for planning a multi-rover collaborative mission in a lunar exploration time network according to an embodiment of the present invention.

[0068] Figure 2 Schematic diagram of obstacle density calculation based on the East China window in an embodiment of the present invention.

[0069] FIG. 3 is a diagram showing optimal movement routes of patrollers corresponding to different numbers of patrollers in an embodiment of the present invention.

[0070] FIG. 4 is a diagram of a lunar road network under different single maximum moving radii in an embodiment of the present invention. DETAILED DESCRIPTION

[0071] In order to enable those skilled in the art to better understand the scheme of the present invention, exemplary implementations or embodiments of the present invention will be described below in conjunction with the accompanying drawings. Obviously, the described implementations or embodiments are only implementations or embodiments of a part of the present invention, not all of them. Based on the implementations or embodiments of the present invention, all other implementations or embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present invention.

[0072] The embodiment of the present invention proposes a multi-rovers collaborative task planning method for a lunar exploration time network, such as Figure 1 As shown, the method comprises the following steps:

[0073] S1. Modeling based on the collaborative exploration mission of multiple lunar rovers;

[0074] S2. Based on the lunar surface collaborative exploration mission, establish a passable road network between the lunar surface detectable points;

[0075] S3. Establish a multi-rovers lunar surface detection time network based on the detection information of the lunar surface traversable road network;

[0076] S4. Use the partheno-genetic algorithm to solve the path of the lunar collaborative exploration mission and obtain the optimal path for each rover to conduct collaborative exploration of the lunar surface.

[0077] The method starts from S1, where a model is built based on the collaborative exploration mission of multiple rovers on the lunar surface.

[0078] According to an embodiment of the present invention, the multi-rover path planning problem is essentially a multi-traveling salesman problem, but takes into account the multiple constraints brought by the lunar mission scenario and the actual lunar environment.

[0079] Each traveling salesman in the classic multi-traveling salesman problem starts from the same mission point. However, for the lunar exploration mission, it is obviously more beneficial for each rover to land at different mission points to shorten the total exploration time. Therefore, the starting point of each rover is not fixed, that is, the starting point of the rover is also one of the optimization contents. For this multi-traveling salesman problem with multiple starting points, inconsistent starting times, and pause time along the way, the problem needs to be converted into a standard multi-traveling salesman problem through the following steps:

[0080] 1) To solve the problem that the starting point of the patroller is not single and uncertain, and does not return to the starting point, a "virtual starting point" v0 can be set. The distance from this virtual starting point to all task points and the time required to reach all task points are both 0, so that all patrollers start from the same virtual starting point. From a practical point of view, this can solve the problem of multiple starting points and unfixed starting points;

[0081] 2) To address the problem of inconsistent departure times, the departure time interval can be added to the detection time of the first mission point, which can be equivalent to assuming that all patrollers depart at the same time.

[0082] Suppose there are m patrollers, one of which sets out at a fixed time T, and needs to detect a total of n mission points. The mission points are at a distance of ten kilometers, and each mission point only needs to be detected once. The detection time of the patroller at the mission point is a fixed value, and the patroller needs to be charged regularly during the journey. In addition, each patroller lands directly at a different mission point and starts the detection mission with the landing point as the starting point. After the detection is completed, it does not need to return to the starting point, and the total detection time is required to be the shortest. The multi-traveling salesman model for the above problem can be described on a weighted undirected graph G = (V, E). Let the set V = {v1, v2, ..., v n} represents the vertex set of graph G, and the set E={ei j =(v i ,v j ):v i ,v j ∈V,i≠j} represents the set of edges of graph G. Using b1, b2, ..., b m represents m patrollers, A={a1,a2,...,a n –1, a n} is a set of task points. Define variables:

[0083]

[0084]

[0085] Formula (1) indicates that when patroller k starts from task point vi and arrives at task point vj, the variable takes the value of 1, otherwise it takes the value of 0; Formula (2) indicates that when patroller k has visited task point i, the variable takes the value of 1, otherwise it takes the value of 0.

[0086] Using variable d ij Represents the task point v i to v j Considering the real lunar environment, the path of the rover cannot be a straight line, so the distance d ij Defined as the length of the path between two task points calculated by the path search algorithm. ij Represents the time from the task point v i Depart to the mission point v j Total time to end, t ij Including the patrol from the task point v i To the mission point v j The time spent on the journey and at the mission point v j The time required to complete the detection is used as the time required to connect the nodes v in the topology graph. i With node v j The edge ij The length of the edge in the weighted undirected graph G is calculated as:

[0087] e ij =t ij =t ij (d ij ,v,t C ,t E ) (3)

[0088] t ij The distance d between two points ij , patrol speed v, detection time t C , Charging time t E Related; The objective function of multi-patrol collaborative detection mission planning is:

[0089] t=min(max(t 1 ,t 2 ,...,t m )) (4)

[0090] in:

[0091]

[0092] The constraints are:

[0093]

[0094] Formula (4) indicates that the total time taken by m patrollers to complete the task is the minimum; Formula (5) indicates the time required for the patroller launched for the kth time to complete the task; Formula (6) indicates that starting from the virtual starting point v0, all detection points are visited by at least one patroller once.

[0095] Then execute S2, in which a passable road network between the detectable points on the lunar surface is established based on the lunar surface collaborative exploration mission.

[0096] According to an embodiment of the present invention, the lunar elevation map is first converted into a binary map of a feasible area. The establishment of the binary map of the feasible area requires extracting terrain information from the lunar DEM map data and calculating the slope, and then treating the map positions with a slope exceeding the patroller's capability limit as obstacles, and treating the positions with a slope less than the patroller's maximum capability limit as feasible areas. Next, a navigable road map between the detection points is established based on the binary map of the feasible area.

[0097] The establishment of the lunar road network consists of two parts: the determination of detection points and the calculation of traversable roads. For the location of the detection points, the requirements of the detection mission and the lunar terrain information can be comprehensively considered to calculate, so that the detection points can be selected outside the obstacle-dense area as much as possible while meeting the detection mission. For the problem of calculating the traversability between the detection points, the A* algorithm is used to search for the path between the detection points on the binary map of the feasible area. If there is a path between the two detection points, it is considered that the two nodes are traversable. The A* algorithm is a classic informed path search algorithm. During the search process, the node with the highest priority is selected from the candidate node list as the next traversal node each time. The heuristic function is an important component of the A* algorithm, and the choice of the heuristic function will have a great impact on the performance of the algorithm. The Manhattan distance is used as the heuristic function to estimate the path cost between the current node and the target node. The Manhattan distance is the sum of the distance between the two nodes in the y-axis direction and the distance in the x-axis direction, which can be expressed as:

[0098] h(n)=|x n -x g |+|y n -y g | (7)

[0099] In the formula, x n ,y n Indicates the coordinates of the current node n; x g ,y g Represents the coordinates of the target node g.

[0100] Considering that the patroller may encounter unavoidable problems caused by unexpected factors when moving between detection points, the patroller needs to return to the previous detection point and re-plan. In order to avoid the serious impact of the retraction operation on the overall execution time of the task. First, the maximum single movement radius r of the patroller is limited according to the driving ability of the patroller, that is, when the straight-line distance between two detection points is greater than r, it is considered that the two points are not directly reachable. If a detection point cannot reach any node within the radius r, it is considered that the detection point is unreachable and is removed from the road network.

[0101] Then, S3 is executed, in which a multi-rover lunar surface detection time network is established based on the lunar surface road network detection information.

[0102] According to an embodiment of the present invention, the connectivity between nodes and path length information can be obtained by using the lunar road network, and the planning goal of multi-patrol collaborative detection is to minimize the detection time. Therefore, before performing detection sequence planning, it is necessary to combine factors such as the patroller's movement speed, charging time, and detection time, and calculate the lunar detection time network based on the lunar road network.

[0103] The travel speed of the rover is affected by many factors such as terrain, geology, and energy. Considering that it is difficult to obtain the contact information between the rover and the lunar surface in advance during a large-scale movement, the present invention mainly considers the impact of terrain and energy on the travel speed of the rover. First, for terrain information, the density of obstacles around the path is used to evaluate the difficulty of the path. Since in the binary map of the feasible area on the lunar surface, the value of the area with obstacles is 1, and the value of the area without obstacles is 0, it can be directly calculated by the binary map. Figure 1 The values ​​in a certain area are summed to evaluate the density of obstacles in this area. To facilitate calculation, the present invention uses a rectangular sliding window to translate on the path to obtain the density of obstacles at each point on the path. The calculation formula for the density of obstacles is:

[0104]

[0105] In the formula, x n ,y n Represents the x, y coordinates of the current path point n; m, n represent the width and height of the sliding window; M(x, y) represents the value at the coordinates of the point (x, y) on the map.

[0106] A trajectory is selected for display, such as Figure 2 As shown in the figure, the box in the figure is the sliding window range for calculating the obstacle density. Assume that the speed of the patroller at full speed is v max , the minimum travel speed is v minThe greater the density of obstacles, the more difficult it is to pass the road. At this time, the speed of the patroller should be reduced. The present invention expresses the relationship between the speed of the patroller and the density of obstacles as follows:

[0107]

[0108] Where b h The obstacle density threshold represents the minimum movement speed of the patroller. When the obstacle density is greater than the threshold, the patroller moves at a constant speed at the minimum speed. l The obstacle density threshold represents the maximum movement speed of the patroller. When the obstacle density is less than the threshold, the patroller moves at a constant speed at the maximum speed.

[0109] After obtaining the patroller's speed, we can calculate the time required for the patroller to complete a path without stopping. The calculation formula is:

[0110]

[0111] In the formula, represents the time required for the patroller to complete the path from detection point i to detection point j without stopping; l represents the patroller movement distance corresponding to a grid in the map; P ij It represents the set of map grid points corresponding to the path from detection point i to detection point j. When node i and node j are not adjacent, the path between the nodes is searched on the established lunar road network through the Dijkstra algorithm.

[0112] In actual lunar exploration missions, it is difficult for the rover to move continuously. Due to energy constraints, the rover needs to stop moving every once in a while to charge with solar energy. The present invention assumes that the rover stops moving to charge after traveling a fixed distance. The calculation formula for the rover's charging time in a path is expressed as:

[0113]

[0114] In the formula, represents the time taken by the patroller to charge on the path from detection point i to detection point j; t c Indicates the time required to fully charge; d ij represents the path length from detection point i to detection point j, d ij =l×size(P ij ), size(P ij ) refers to the path P ij The number of grid points; d c Indicates the distance the rover travels between charges.

[0115] When the patroller arrives at the detection point, it is assumed that the patroller immediately starts the detection operation, and the operation time at each detection point is set to It is assumed that the detector will be fully charged after the detection and before moving to the next detection point, and the charging time is included in the detection time. Also included in t ij After comprehensively considering the travel speed, charging time, and detection time, the total time it takes for the detector to travel from detection point i to detection point j is expressed as:

[0116]

[0117] Then, S4 is executed. In S4, a partheno-genetic algorithm is used to solve the path planning of the lunar collaborative exploration mission, and the optimal path for each rover to perform the lunar collaborative exploration is obtained.

[0118] According to an embodiment of the present invention, the Partheno-Genetic Algorithm (PGA) is an improved genetic algorithm. It no longer performs the crossover operation of the parent chromosomes in the traditional genetic algorithm, but performs operations such as gene inversion, gene recombination, and gene translocation on the parent chromosomes. It is simple to operate and has a high search efficiency, but its performance is sensitive to parameters and it is difficult to obtain suitable parameters. It has been improved by combining the selection operator with the mutation operator and removing the mutation probability parameter, thereby obtaining a more efficient and stable performance. Therefore, this algorithm is selected to solve the problem of lunar collaborative exploration.

[0119] (1) Coding

[0120] The present invention adopts a two-segment chromosome encoding method to express the solution, and encodes n lunar surface detection nodes and m patrollers into a chromosome consisting of two parts, the first segment is a chromosome with a length of n-1, wherein the n lunar surface detection nodes are represented by 1, 2..., n real numbers, representing that the patrollers start from the same starting point and have remaining n-1 detection nodes to be traversed; the second segment is m-1 positive integers, and the n lunar surface detection nodes are divided into m segments, which respectively represent the lunar surface detection nodes to be traversed by the m patrollers, and their values ​​are the locations where the chromosome breaks and the breakpoint values ​​will determine the number of nodes that each patroller needs to detect. Therefore, the numbers in the second segment need to be arranged in ascending order from small to large, and the difference between adjacent numbers should not be less than the minimum number of nodes detected by the patroller.

[0121] (2) Population initialization

[0122] The initial population of the genetic algorithm is the basis for subsequent iterative evolution. The wider its coverage, the more likely it is to contain the global optimal solution. In the present invention, there are n lunar exploration nodes, and the lunar exploration nodes are numbered from 1 to n. Excluding the starting node 1, individuals with randomly generated chromosome genes arranged randomly from 2 to n and n - 1 breakpoints are generated. Here, the arrangement is the visiting order of each rover.

[0123] (3) Fitness function

[0124] The collaborative detection objective function is used as the fitness function of the genetic algorithm. The path with the highest fitness is the one with the least total time for all rovers to complete all lunar explorations.

[0125] (4) Selection operation

[0126] Randomly select h individuals from the current population, calculate their fitness according to the above fitness function, and store the individual with the highest fitness in a temporary population containing h individuals. The other h - 1 individuals are removed and replaced by the individual with the highest fitness through mutation operation.

[0127] (5) Mutation operation

[0128] Randomly generate two mutation positions I and J (I < J), and the insertion position P of the mutation segment. Each individual in the aforementioned temporary population is mutated in different ways. The ways include: no operation, gene inversion, gene transposition, gene left shift, gene right shift, and only modifying the breakpoint of the latter part without modifying the former part of the individual. Among them, resetting the breakpoint can be used in combination with gene inversion, gene transposition, and gene shift respectively.

[0129] Gene inversion operation: Reverse the gene order from position I to position J, and insert the gene segment between I and J as a whole at position P;

[0130] Gene transposition operation: Exchange the genes at positions I and J, and insert the gene segment between I and J as a whole at position P;

[0131] Gene shift operation: Move the gene segment from position I + 1 to position J as a whole one position to the left, and then place the gene at position I at position J; or move the gene segment from position I to J - 1 as a whole one position to the right, and then place the gene at position J at position I. Finally, insert the gene segment between I and J as a whole at position P.

[0132] (6) Iteration termination

[0133] When the set termination condition is met, such as reaching the set maximum number of generations, stop the iterative search. At this time, the solution is the optimal path for each rover to conduct collaborative detection.

[0134] The technical effect of the present invention is further verified through experiments.

[0135] First, according to the detection time of different departure time intervals and the number of rovers in the lunar collaborative detection mission, the optimal allocation plan of the lunar detection time network is obtained. In the lunar multi-rovers joint detection mission, the number of rovers and the departure time interval between rovers will have an important impact on the final detection time. The departure time interval of the rovers is determined by the landing time and preparation time of each rover. Assuming that the departure time interval between the rovers is the same, the departure time intervals of 10h, 20h, 30h, and 40h are set for comparison. The number of rovers is set to 3, and the maximum single movement radius r of the rovers is set to 10km.

[0136] As the departure time interval increases, the maximum detection time of the three patrollers also increases approximately linearly, and the detection time of each patroller under different departure time intervals. The time of each detector includes the departure time interval, that is, the time of each detection is the time from the departure of the first detector to the end of the detection of each detector. From the simulation results, it can be obtained that the patrollers that depart late will be automatically assigned fewer detection tasks, and the time of each patroller can be balanced so that the three patrollers can end the task together at a similar time.

[0137] The number of patrollers was compared and analyzed, and the number of patrollers was set to 2, 3, 4, and 5 for comparison. The departure time interval between patrollers was set to 10h, and the single maximum movement radius r of the patroller was set to 10km. Each case was simulated 5 times repeatedly. Figure 3 is the optimal movement route map of the patrollers corresponding to different numbers of patrollers of the present invention, where Figure 3 (a) is the optimal movement route of the patrollers corresponding to the number of patrollers of 2, Figure 3 (b) is the optimal movement route of the patrollers corresponding to the number of patrollers of 3, Figure 3 (c) is the optimal movement route of the patrollers corresponding to the number of patrollers of 4, and Figure 3 (d) is the optimal movement route of the patrollers corresponding to the number of patrollers of 5. The average value and optimal value of the 5 times can be seen from the calculation results. Under the condition of the same number of detection nodes, as the number of patrollers increases, the task volume of each patroller decreases, and the overall detection time is also significantly reduced. The calculation results of the algorithm are in line with expectations.

[0138] Then, according to the road network connectivity status of the maximum single moving radius of multiple patrollers under the lunar collaborative exploration mission, the optimal path planning scheme is obtained. In the process of constructing the lunar road network, the maximum single moving radius of the patroller is a parameter that determines the density of the road network. When the patroller encounters an impassable situation during the journey, it needs to fall back to the previous detection node to re-plan the detection sequence. By setting the maximum single moving radius, the fallback operation can be prevented from having a serious impact on the overall execution time of the task and reduce the risk of the exploration mission. The selection of the maximum single moving radius is compared, and it is set to 10km, 20km, and 30km respectively. The lunar road networks corresponding to the three cases are shown in Figure 4. It can be seen that the connectivity of the road network is significantly improved with the increase of the maximum single moving radius.

[0139] As the maximum single movement radius increases, the connectivity of the lunar road network improves, giving the rover more optional paths, thus reducing the overall detection time to a certain extent. When the maximum single movement radius is 20km, the average total detection time is reduced by 2.1% compared to the radius of 10km. When the maximum single movement radius is 30km, the average total detection time is reduced by 3.0% compared to the radius of 10km. It can be seen that the improvement of the connectivity of the lunar road network has a certain effect on reducing the total detection time, but the effect is limited. When setting the maximum single movement radius, it is necessary to combine the capabilities of the rover to balance the mission risk and detection time.

[0140] In summary, the present invention proposes a multi-rover collaborative task planning method for the lunar exploration time network, which improves the connectivity of the lunar road network and provides more optional paths for the rovers, thereby reducing the overall exploration time to a certain extent and providing the best path solution.

[0141] Another embodiment of the present invention provides a multi-rover collaborative mission planning system for a lunar exploration time network, the system comprising:

[0142] A model building module configured to model a multi-rovers collaborative exploration mission on the lunar surface;

[0143] A traversable road network establishment module, which is configured to establish a traversable road network between traversable points on the lunar surface based on the lunar surface collaborative exploration mission;

[0144] A detection time network establishment module, which is configured to establish a multi-rover lunar surface detection time network based on the detection information of the lunar surface traversable road network;

[0145] The optimal path calculation module is configured to perform path planning for the lunar collaborative exploration mission based on a partheno-genetic algorithm to obtain the optimal path for each rover to perform lunar collaborative exploration.

[0146] The functions of the multi-rover collaborative task planning system of the lunar surface exploration time network described in this embodiment can be explained by the multi-rover collaborative task planning method of the lunar surface exploration time network mentioned above. Therefore, for the parts not described in detail in this embodiment, please refer to the above method embodiments and will not be repeated here.

[0147] Although the present invention has been described according to a limited number of embodiments, it will be apparent to those skilled in the art, with the benefit of the above description, that other embodiments are contemplated within the scope of the invention thus described. The disclosure of the present invention is intended to be illustrative rather than restrictive of the scope of the invention, which is defined by the appended claims.

Claims

1. A multi-rovers collaborative mission planning method for lunar exploration time network, characterized in that: include: S1. Modeling based on the collaborative exploration mission of multiple lunar rovers; The specific steps include: Suppose there are m rovers on the moon, one of which sets out at a fixed time T, and needs to detect n mission points in total. Each mission point only needs to be detected once. The detection time of the rover at the mission point is fixed, and the rover needs to be charged regularly during the journey; each rover lands directly at a different mission point and starts the detection mission with the landing point as the starting point. After the detection is completed, it does not need to return to the starting point, and the total detection time is required to be the shortest. The multi-traveling salesman model of the above problem is described on a weighted undirected graph G = (V, E), let the set V = {v1, v2, ..., v n } represents the vertex set of graph G, and the set E = {e ij =(v i ,v j ):v i ,v j ∈V,i≠j} represents the set of edges of graph G, using b1, b2, ..., b m represents m patrollers, A={a1,a2,…,a n-1 ,a n } is a set of task points; define the following variables: Formula (1) indicates that when patroller k moves from task point v i Departure, arrive at the mission point v j When , the variable takes the value of 1, otherwise it takes the value of 0; Formula (2) means that when patroller k has visited task point i, the variable takes the value of 1, otherwise it takes the value of 0; Using variable d ij Represents the task point v i to v j The distance between ij It is defined as the length of the path between two task points calculated by the path search algorithm; t ij Represents the time from the task point v i Depart to the mission point v j Total time to end, t ij Including the patroller from the task point v i To the mission point v j The time spent on the journey and at the mission point v j The time required to complete the detection is used as the time required to connect the nodes v in the topology graph. i With node v j The edge ij The length of the edge in the weighted undirected graph G is calculated as: e ij =t ij =t ij (d ij ,v,t C ,t E ) (3) t ij The distance d between two points ij , patrol speed v, detection time t c , Charging time t E Related; The objective function of multi-patrol collaborative detection mission planning is: t=min(max(t 1 ,t 2 ,...,t m )) (4) in: The constraints are: Formula (4) indicates that the total time taken by m patrollers to complete the task is the minimum; Formula (5) indicates the time required for the patroller launched for the kth time to complete the task; Formula (6) indicates that starting from the virtual starting point v0, all detection points are visited by at least one patroller once; S2. Based on the lunar surface collaborative exploration mission, establish a passable road network between the lunar surface detectable points; the specific steps include: First, the lunar elevation map is converted into a binary map of the feasible area. The establishment of the binary map of the feasible area requires extracting terrain information from the lunar elevation map and calculating the slope. Then, the map positions with slopes exceeding the rover's capacity limit are regarded as obstacles, and the positions with slopes less than the rover's maximum capacity limit are regarded as feasible areas. Next, a drivable road map between detection points is established based on the binary map of the feasible area, including determining detection points and calculating drivable roads; S3. Establishing a multi-rovers lunar surface detection time network based on the detection information of the lunar surface traversable road network. The specific steps include: First, for terrain information, the density of obstacles around the path is used to evaluate the difficulty of the path. A rectangular sliding window is used to translate on the path to obtain the density of obstacles at each point on the path. The relationship between the patrol vehicle's speed and obstacle density is expressed as: Where b h The obstacle density threshold represents the minimum movement speed of the patroller. When the obstacle density is greater than the threshold, the patroller moves at a constant speed at the minimum speed. l The obstacle density threshold represents the maximum movement speed of the patroller. When the obstacle density is less than the threshold, the patroller moves at a constant speed at the maximum speed. After obtaining the patroller's speed, calculate the time required for the patroller to complete a path without stopping. The calculation formula is: In the formula, represents the time required for the patroller to complete the path from detection point i to detection point j without stopping; l represents the patroller movement distance corresponding to a grid in the map; P ij Represents the set of map grid points corresponding to the path from detection point i to detection point j. When node i and node j are not adjacent, the path between the nodes is searched on the established lunar road network by Dijkstra algorithm; The calculation formula for the patroller's charging time in a path is expressed as: In the formula, represents the time taken by the patroller to charge on the path from detection point i to detection point j; d ij represents the path length from detection point i to detection point j, d ij =l×size(P ij ), size(P ij ) refers to the path P ij The number of grid points; d c Indicates the distance the patroller travels between two charges; When the patroller arrives at the detection point, it is assumed that the patroller immediately starts the detection operation, and the operation time at each detection point is set to Assuming that the detector is fully charged after detection and before moving to the next detection point, and the charging time is included in the detection time; the total time taken by the detector from detection point i to detection point j is expressed as: S4. Use the partheno-genetic algorithm to solve the path of the lunar collaborative exploration mission and obtain the optimal path for each rover to conduct collaborative exploration of the lunar surface.

2. The method for planning multi-rovers collaborative missions in a lunar exploration time network according to claim 1, characterized in that: The specific steps of determining detection points and calculating drivable roads in S2 include: For the location of the detection point, the requirements of the detection mission and the lunar terrain information are taken into consideration, and the detection point is selected outside the obstacle-dense area while meeting the detection mission; For the problem of calculating the feasibility between detection points, the A* algorithm is used to search for paths between detection points on the binary map of the feasible area. If there is a path between two detection points, it is considered that the two nodes are passable. During the search process, the node with the highest priority is selected from the candidate node list as the next traversal node each time; Manhattan distance is used as the heuristic function in the A* algorithm to estimate the path cost between the current node and the target node. Manhattan distance is the sum of the distance between the two nodes in the y-axis direction and the distance in the x-axis direction.

3. The method for planning multi-rovers collaborative tasks in a lunar exploration time network according to claim 2, characterized in that: The specific steps of determining the detection point in S2 also include: The maximum single moving radius r of the patroller is limited according to its driving capability. That is, when the straight-line distance between two detection points is greater than r, it is considered that the two points are not directly reachable. If a detection point cannot reach any node within the radius r, it is considered that the detection point is unreachable and is removed from the road network.

4. The method for multi-rovers collaborative mission planning of a lunar exploration time network according to claim 3 is characterized in that: The calculation formula for obstacle density in S3 is: In the formula, x n ,y n Indicates the x, y coordinates of the current path point n; m, n represent the width and height of the sliding window; M(x, y) represents the value at the coordinates of the point (x, y) on the map.

5. The method for multi-rovers collaborative mission planning of lunar exploration time network according to claim 4 is characterized in that: The specific steps of S4 include: Coding: The expression of the solution adopts a two - segment chromosome coding method. The n lunar surface exploration nodes and m rovers are encoded into a chromosome consisting of two parts. The first segment is a chromosome with a length of n - 1, where the n lunar surface exploration nodes are represented by n real numbers 1, 2, …, n, representing that the rover starts from the same starting point and there are n - 1 remaining exploration nodes to traverse. The second segment is m - 1 positive integers, which divide the n lunar surface exploration nodes into m segments, respectively representing the lunar surface exploration nodes to be traversed by m rovers. Their values are the positions where the chromosome is broken. The value of the breakpoint determines the number of nodes that each rover needs to detect. The numbers in the second segment are arranged in ascending order from small to large, and the difference between adjacent numbers is not less than the minimum number of nodes detected by the rover. Population initialization: There are n lunar surface exploration nodes, so the lunar surface exploration nodes are numbered from 1 to n. Excluding the starting node 1, individuals with randomly generated chromosome genes randomly arranged from 2 to n and n - 1 breakpoints are generated. Here, the arrangement is the visiting order of each rover. Fitness function: The collaborative detection objective function is used as the fitness function of the genetic algorithm. The path with the highest fitness is the one where all rovers complete all lunar surface detections in the least total time. Selection operation: Randomly select h individuals from the current population, calculate their fitness according to the above - mentioned fitness function, and store the individual with the highest fitness in a temporary population containing h individuals. The other h - 1 individuals are removed and replaced by the individual with the highest fitness through the mutation operation. Mutation operation: Randomly generate two mutation positions I and J, where I < J, and the insertion position P of the mutation segment. Different methods are used to mutate each individual in the aforementioned temporary population. Iteration termination: When the set termination condition is met, the iterative search stops. At this time, the solution is the optimal path for each rover to conduct collaborative detection.

6. The method for multi-rovers collaborative mission planning of a lunar exploration time network according to claim 5 is characterized in that: The different methods of mutating each individual in the aforementioned temporary population using different methods include: not performing any operation, gene inversion, gene transposition, gene left - shift, gene right - shift, and only modifying the breakpoints of the second part without modifying the first part of the individual. Among them, resetting the breakpoints is used in combination with gene inversion, gene transposition, and gene shift respectively.

7. A multi-rovers collaborative mission planning system for lunar exploration time network, characterized in that: Including: Model establishment module, which is configured to build a model according to the lunar surface multi - rover collaborative detection task. The specific steps include: Suppose there are m rovers on the moon, one of which sets out at a fixed time T, and needs to detect n mission points in total. Each mission point only needs to be detected once. The detection time of the rover at the mission point is fixed, and the rover needs to be charged regularly during the journey; each rover lands directly at a different mission point and starts the detection mission with the landing point as the starting point. After the detection is completed, it does not need to return to the starting point, and the total detection time is required to be the shortest. The multi-traveling salesman model of the above problem is described on a weighted undirected graph G = (V, E), let the set V = {v1, v2, ..., v n } represents the vertex set of graph G, and the set E = {e ij =(v i ,v j ):v i ,v j ∈V,i≠j} represents the set of edges of graph G, using b1, b2, ..., b m represents m patrollers, A={a1,a2,…,a n-1 ,a n } is a set of task points; define the following variables: Formula (1) indicates that when patroller k moves from task point v i Departure, arrive at the mission point v j When , the variable takes the value of 1, otherwise it takes the value of 0; Formula (2) means that when patroller k has visited task point i, the variable takes the value of 1, otherwise it takes the value of 0; Using variable d ij Represents the task point v i to v j The distance between ij It is defined as the length of the path between two task points calculated by the path search algorithm; t ij Represents the time from the task point v i Depart to the mission point v j Total time to end, t ij Including the patrol from the task point v i To the mission point v j The time spent on the journey and at the mission point v j The time required to complete the detection is used as the time required to connect the nodes v in the topology graph. i With node v j The edge ij The length of the edge in the weighted undirected graph G is calculated as: e ij =t ij =t ij (d ij ,v,t C ,t E ) (3) t ij The distance d between two points ij , patrol speed v, detection time t c , Charging time t E Related; The objective function of multi-patrol collaborative detection mission planning is: t=min(max(t 1 ,t 2 ,...,t m )) (4) Among them: The constraint conditions are: Equation (4) represents that the total time spent by m rovers to complete the task is the minimum; Equation (5) represents the time required for the rover launched at the k - th time to complete the task; Equation (6) represents that starting from the virtual starting point v0, all detection points are visited at least once by one rover. Passable road network establishment module, which is configured to establish a passable road network between detectable points on the lunar surface based on the lunar surface collaborative detection task. The specific steps include: First, the lunar elevation map is converted into a binary map of the feasible area. The establishment of the binary map of the feasible area requires extracting terrain information from the lunar elevation map and calculating the slope. Then, the map positions with slopes exceeding the rover's capacity limit are regarded as obstacles, and the positions with slopes less than the rover's maximum capacity limit are regarded as feasible areas. Next, a drivable road map between detection points is established based on the binary map of the feasible area, including determining detection points and calculating drivable roads; The detection time network establishment module is configured to establish a multi-rover lunar surface detection time network based on the detection information of the lunar surface traversable road network; the specific steps include: First, for terrain information, the density of obstacles around the path is used to evaluate the difficulty of the path. A rectangular sliding window is used to translate on the path to obtain the density of obstacles at each point on the path. The relationship between the patrol vehicle's speed and obstacle density is expressed as: Where b h The obstacle density threshold represents the minimum movement speed of the patroller. When the obstacle density is greater than the threshold, the patroller moves at a constant speed at the minimum speed. l The obstacle density threshold represents the maximum movement speed of the patroller. When the obstacle density is less than the threshold, the patroller moves at a constant speed at the maximum speed. After obtaining the patroller's speed, calculate the time required for the patroller to complete a path without stopping. The calculation formula is: In the formula, represents the time required for the patroller to complete the path from detection point i to detection point j without stopping; l represents the patroller movement distance corresponding to a grid in the map; P ij Represents the set of map grid points corresponding to the path from detection point i to detection point j. When node i and node j are not adjacent, the path between the nodes is searched on the established lunar road network by Dijkstra algorithm; The calculation formula for the patroller's charging time in a path is expressed as: In the formula, represents the time taken by the patroller to charge on the path from detection point i to detection point j; d ij represents the path length from detection point i to detection point j, d ij =l×size(P ij ), size(P ij ) refers to the path P ij The number of grid points; d c Indicates the distance the patroller travels between two charges; When the patroller arrives at the detection point, it is assumed that the patroller immediately starts the detection operation, and the operation time at each detection point is set to Assuming that the detector is fully charged after detection and before moving to the next detection point, and the charging time is included in the detection time; the total time taken by the detector from detection point i to detection point j is expressed as: The optimal path calculation module is configured to perform path planning for the lunar collaborative exploration mission based on a partheno-genetic algorithm to obtain the optimal path for each rover to perform lunar collaborative exploration.

Images