Robot autonomous navigation algorithm and storage medium under adversarial environment

Abstract

The application discloses a robot autonomous navigation algorithm in an environment, and a storage medium, which comprises the following steps: firstly, constructing a navigation tactical cost map containing enemy firepower threat information based on a terrain shelter judgment algorithm, and optimizing and calculating an optimal navigation path based on the navigation tactical cost map, and finally executing a tactical navigation action capable of guaranteeing safety to the maximum; further constructing a dynamic tactical action planning algorithm based on a tactical benefit analysis, and realizing a close attack action on a specified enemy target and a retreat self-protection instruction. The application can evade obstacles while reducing exposure under enemy firepower, and realize tactical navigation actions of attack and retreat. The application focuses on path planning and map construction of a military robot in a shooting confrontation environment, so that the robot can evade obstacles while having the functions of tactical navigation, attack and retreat according to tactical information such as enemy firepower distribution.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot autonomous navigation, and particularly relates to a robot autonomous navigation algorithm in an adversarial environment and a storage medium. BACKGROUND

[0002] With the continuous maturity of robot application technology, military robots are also developing towards autonomy and intelligence. However, when the military robots enter the real battlefield environment and shoot against the enemy, the traditional navigation planning algorithm often cannot adapt to the special needs in the battlefield confrontation environment in time.

[0003] Robot navigation refers to the ability of a robot to determine its own position in a reference frame and then plan a path to a certain target position. In order to navigate in its environment, a robot or any other mobile device needs a representation, i.e. an environmental map, and the ability to interpret this representation. Navigation can be defined as a combination of three basic abilities: self-localization, path planning, map construction and interpretation. Robot localization represents the ability of a robot to establish its own position and orientation within a reference frame. Path planning is actually an extension of localization, because it requires determining the position of the robot's current position and target position, both within the same reference frame or coordinate system. Map construction can take the form of a metric map or any symbol representing a position in the robot's reference frame.

[0004] The existing navigation planning algorithm usually represents the through cost corresponding to different positions in the current environment by constantly updating a cost map calculated by obstacle collision danger, and runs a path planning algorithm such as Dijkstra or A on it to calculate the running track of the robot.

[0005] However, when navigating in a complex battlefield enemy confrontation environment, in addition to the basic obstacle blockage, complex battlefield information such as enemy position and fire threat distribution also determines whether the robot navigation planning is safe. At the same time, the attack and retreat actions in combat also put forward higher requirements for the navigation algorithm. Therefore, the traditional navigation algorithm cannot consider the fire threat brought by the enemy in the actual confrontation environment, thereby reducing the safety in the robot navigation process, and the reliability and diversity of the navigation task. SUMMARY

[0006] The robot autonomous navigation algorithm in an adversarial environment provided by the present application, i.e. a navigation planning algorithm for autonomous movement of a military robot in a shooting confrontation environment, is used to solve the above technical problems, so as to reduce the exposure degree under the enemy fire while avoiding obstacles, and realize the tactical navigation action of attacking and retreating to the target.

[0007] To achieve the above purpose, the present application adopts the following technical solutions:

[0008] A robot autonomous navigation algorithm in the face of the environment, comprising the following steps,

[0009] Firstly, a navigation tactical cost map containing enemy firepower threat information is constructed based on a terrain occlusion determination algorithm, and an optimal navigation path is calculated based on the navigation tactical cost map, and finally a tactical navigation action is executed which can maximize the safety of the robot;

[0010] Thereafter, a dynamic tactical action planning algorithm based on tactical benefit analysis is further constructed to realize the approach attack action on the specified enemy target and the retreat self-protection instruction.

[0011] Further, the tactical cost map is a two-dimensional grid map with the same size as the current map, and the value on each grid represents the cost of the corresponding robot passing through the position;

[0012] The tactical cost map is stacked by several cost map layers, each cost map layer represents different moving cost information, and the complete tactical cost map can be obtained by stacking and adding all cost map layers.

[0013] Further, the tactical cost map includes four cost map layers stacked by the enemy firepower threat layer, the static map layer, the obstacle layer and the inflation layer;

[0014] Enemy firepower threat layer: used to represent the spatial distribution of enemy firepower threat;

[0015] Traverse each enemy unit, and increase the corresponding firepower threat value of all grid points on the layer without obstacle occlusion between the unit according to the inverse square relationship between enemy firepower weight and distance; Specifically, when the number of enemies is K, for an m × n grid cost map layer, the firepower cost J ij of grid point (i, j) is calculated as follows:

[0016] (1)

[0017] (2)

[0018] Wherein, is the firepower cost value caused by the k th enemy at grid point i, j , Q k is the firepower weight of the k th enemy, which is proportional to the firepower intensity of the enemy; is the distance from the th enemy to grid point .

[0019] Static map layer: used to represent the known ground obstacles to the navigation movement cost;

[0020] The current environment two-dimensional obstacle map input before is converted into a cost layer, if the grid is occupied by obstacles, the cost of the point is set to the preset collision cost, otherwise the cost is zero;

[0021] Dynamic obstacle layer: used to represent the laser radar scanned obstacles to the navigation movement cost;

[0022] The laser radar scanned obstacle data points will be stored in a queue, and the cost of the obstacle data point is set to the collision cost;

[0023] At the same time, the queue is refreshed regularly to remove old scanning data, so as to ensure the real-time of the cost information of dynamic obstacles;

[0024] Inflation layer: used to represent the movement risk caused by too close to obstacles;

[0025] The inflation cost layer expands the collision cost part in the static map layer and the dynamic obstacle layer to a certain distance in all directions, and adds a cost decay area, the specific implementation formula of the inflation layer is as follows:

[0026] (3)

[0027] Where J ij is the inflation layer cost at grid point (i, j), d ij is the distance from grid point (i, j) to the nearest obstacle, r is the set inflation radius, and J0 is the cost value corresponding to the obstacle.

[0028] Further, the dynamic tactical action planning algorithm comprises:

[0029] On the established tactical cost map layer, run the A* path planning algorithm to realize the robot tactical pathfinding effect; in the tactical pathfinding A* planning algorithm, the cost n of node f ( n ) formula is as follows:

[0030] f ( n ) = g ( n ) + h ( n )(4)

[0031] Wherein, g ( n) is the movement cost of the robot from the start point to the node n The movement cost of the robot from the start point to the node

[0032] (5)

[0033] wherein, d k-1,k is the movement distance cost of the robot from the start point to the node k - 1 k , and in addition, the accumulated value of the enemy fire cost on the path is added to the original A* algorithm , and the weight parameter K g is used to adjust the weight of the enemy fire threat in the navigation planning calculation;

[0034] The heuristic cost function g ( n ) selects the Manhattan distance from the node n to the target point as the prediction approximation to reduce the planning algorithm calculation overhead, and the calculation formula is as follows:

[0035] (6)

[0036] wherein ( x n , y n ), ( x t , y t ) are the coordinates of the node n and the target point t in the map grid.

[0037] Further, the dynamic tactical action planning algorithm further includes autonomous tactical movement planning, specifically including close attack planning:

[0038] The close attack planning includes specifying the enemy unit to be attacked or giving its position coordinates, and the robot will autonomously calculate an attack position of the enemy unit that can be shot by the on-board direct-fire weapon as a navigation planning target point, and plan a movement path to the target point through the above navigation algorithm;

[0039] The calculation method of the attack position is as follows: in a square region with the enemy unit coordinates ( x 0 , y 0) as the center and a side length of 2 d , the benefit value of the attack position coordinates ( x, y ) is calculated by the following cost formula, and the coordinate with the maximum benefit value is obtained as the attack position through optimization.

[0040] (7)

[0041] wherein, α ( x, y ) is an unobstructed judgment function, which is 1 when there is no obstacle between the attack position coordinate ( x, y ) and the enemy unit coordinate ( x 0 , y 0), otherwise it is 0;

[0042] R ( x, y ) is the attack position's benefit value, which is determined by the distance between the attack position and the enemy unit and the ideal attack distance ; the closer the attack launch distance is to the ideal distance, the higher the benefit is; the function for calculating the benefit value is:

[0043] (8)

[0044] K q ·Q ( x, y ) is the movement cost of the robot to reach the attack position, which is the geometric distance between the current position of the robot ( x r , y r ) and the attack position ( x 0 , y 0), which is an approximation of the actual movement distance; the parameter K q is used to adjust the weight of the movement cost in the optimization calculation;

[0045] K f · J (x,y) is the enemy fire threat cost of the attack position, and the parameter K f is used to adjust the weight of the fire threat cost in the optimization calculation.

[0046] Further, the autonomous tactical movement planning also includes a retreat self-protection planning, when the robot's own condition is in danger, the robot can adopt a retreat self-protection movement strategy, and according to the position distribution and fire threat distribution of the enemy, autonomously move to a safer position; the optimization formula for finding the retreat self-protection movement position is as follows, the ideal retreat position ( x, y ) can be obtained by optimizing the retreat position cost:

[0047] (9)​

[0048] Q x, y ) and J x, y ) are respectively the movement cost of the robot reaching the retreat point and the enemy firepower threat value of the point.

[0049] In another aspect, a computer readable storage medium stores a computer program, which, when executed by a processor, causes the processor to perform the steps of the above method.

[0050] From the above technical solution, the autonomous navigation algorithm of the robot in the confrontation environment of the application is used to solve the technical problem of battlefield confrontation environment navigation, so that the exposure degree under the enemy firepower can be reduced while avoiding obstacles, and tactical navigation actions of attacking and retreating to the target can be realized.

[0051] The application focuses on the path planning and map construction of the military robot in the shooting confrontation environment, so that the robot can realize the functions of tactical navigation, attack and retreat according to the tactical information such as enemy firepower distribution while avoiding obstacles.

[0052] The application focuses on the path planning and map construction of the military robot in the shooting confrontation environment, so that the robot can realize the functions of tactical navigation, attack and retreat according to the tactical information such as enemy firepower distribution while avoiding obstacles.

[0053] In general, the application focuses on the path planning and map construction of the military robot in the shooting confrontation environment, so that the robot can realize the functions of tactical navigation, attack and retreat according to the tactical information such as enemy firepower distribution while avoiding obstacles.

[0054] The application has the advantages of:

[0055] 1. A tactical cost map construction method for describing the spatial distribution of enemy firepower threats is proposed, so that the navigation and mapping link of the robot can reflect the quantitative spatial representation of the enemy and our situation.

[0056] 2. A navigation planning algorithm for autonomous movement of the military robot in the shooting confrontation environment is proposed, which can integrate threat information from enemy attacks into the optimization calculation of the navigation algorithm, thereby improving the survival ability and task reliability of the robot in the confrontation environment.

[0057] 3. Based on the above algorithm, an autonomous navigation algorithm for the military robot to attack the target or retreat for self-protection is further proposed, thereby realizing the specific tactical role of the military robot navigation algorithm and enhancing the diversity of algorithm functions. ​​BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 is a flow chart of the method of the present application;

[0059] Figure 2 is a structural block diagram of the algorithm of the present application;

[0060] Figure 3 is a tactical cost map of the present application describing the fire threat of the enemy robot No. 1, the right side is a schematic diagram of the simulated confrontation environment, and the left side is a cost map for navigation of the robot No. 2 of our side, the gray radial pattern on the left side is the spatial distribution of the cost of the fire threat of the enemy, and the brighter the color is, the higher the threat cost is;

[0061] Fig. 4 is a schematic diagram of the tactical path planning of the robot No. 2 under the fire threat of the enemy robot No. 1, the black straight line in the right diagram is the movement target indication line thereof, and the black broken line in the left diagram is the tactical movement path calculated thereby, which is selected to bypass from behind the shelter to avoid the fire threat of the enemy robot on the left side;

[0062] Figure 5 is a schematic diagram of the tactical path planning of the robot No. 2 not under the fire threat of the enemy robot No. 1, the black straight line in the right diagram is the movement target indication line thereof, and the broken line in the left diagram is the tactical movement path calculated thereby, which directly passes through the safe area;

[0063] Fig. 6 is the attack position planning of the robot No. 1 approaching the right side robot No. 2, and the circle is the ideal attack point calculated by optimization;

[0064] Figure 7 is the retreat self-protection planning of the robot No. 1, and the left straight line lower end point is the ideal retreat point planned thereby, which is located behind the shelter that cannot be shot by the enemy robot No. 2. DETAILED DESCRIPTION

[0065] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application.

[0066] The algorithm of the embodiments of the present application needs to input a completely or partially constructed current environment two-dimensional map for representing the passable area and obstacle blocking information. Meanwhile, the position information of the perceived enemy units needs to be input for subsequent calculation of the spatial distribution of the enemy fire threat. Finally, the laser radar sensor and odometer information of itself needs to be input for detection of dynamic obstacles and real-time positioning of itself in the map.

[0067] The algorithm in this embodiment of the invention will first construct a navigation tactical cost map containing enemy fire threat information based on the terrain occlusion determination algorithm, and then optimize and calculate the optimal navigation path based on it, and finally execute tactical navigation actions that can maximize the safety of itself.

[0068] Subsequently, a dynamic tactical action planning algorithm based on tactical benefit analysis was further constructed to realize close-range attack actions against designated enemy targets, as well as retreat and self-defense commands.

[0069] The following sections will explain each of the key technical aspects of the solution.

[0070] like Figure 1 As shown, the robot autonomous navigation algorithm in an adversarial environment described in this embodiment includes the following steps:

[0071] 1. Tactical Cost Map

[0072] First, a navigation tactical cost map containing enemy fire threat information is constructed. The tactical cost map is a two-dimensional grid map of the same size as the current map, where the value on each grid represents the cost for the corresponding robot to move to that location. The tactical cost map is composed of several stacked cost layers, each representing different movement cost information. By stacking and summing all the cost layers, the complete tactical cost map is obtained.

[0073] The tactical cost map is composed of four cost layers stacked together: an enemy firepower threat layer, a static map layer, an obstacle layer, and an expansion layer. The methods for obtaining it are described below.

[0074] (1) Enemy fire threat layer: used to represent the spatial distribution of enemy fire threat.

[0075] For each enemy unit, for all unobstructed grid points on the layer that are not occluded by obstacles, increase the corresponding fire threat value based on the enemy's firepower weight and the inverse square law of distance. Specifically, when the number of enemies is K, for an m × n grid cost layer, the firepower cost J of grid point (i, j) is... ij The calculation formula is as follows:

[0076] (1)

[0077] (2)

[0078] in, It is the first k One enemy at the grid point ( i, j The numerical value of the firepower cost caused. Q k For the first kThe firepower weight of each enemy is directly proportional to the firepower intensity of that enemy. d ij For the first k One enemy to the grid point ( i, j The distance.

[0079] (2) Static map layer: used to represent the movement cost caused by known obstacles on the ground to navigation.

[0080] Simply convert the previously input 2D obstacle map of the current environment into a cost layer. If a grid point is occupied by an obstacle, set the cost of that point to the preset collision cost; otherwise, set the cost to zero.

[0081] (3) Dynamic obstacle layer: used to represent the movement cost of navigation caused by obstacles scanned by the lidar.

[0082] The obstacle data points scanned by the lidar will be stored in a queue, and the cost of the obstacle data points will be set as the collision cost.

[0083] Meanwhile, the queue is periodically refreshed to remove old scan data, ensuring that the cost information of dynamic obstacles is real-time.

[0084] (4) Inflated layer: used to represent the risk of movement caused by being too close to an obstacle.

[0085] Because the robot has physical dimensions, there is still a risk of collision even when there is a distance between the robot's coordinate system center and the obstacle. Furthermore, the robot should stay as far away from the obstacle as possible to avoid collisions caused by inaccurate motion control when too close. Therefore, the dilation cost layer expands the collision cost portion of (2) the static map layer and (3) the dynamic obstacle layer by a certain distance in all directions, and adds a cost attenuation area. The specific implementation formula for the dilation layer is as follows:

[0086] (3)

[0087] Among them, J ij For the cost of the dilation layer at grid point (i, j), d ij is the distance from grid point (i, j) to the nearest obstacle, r is the set expansion radius, and J0 is the cost value corresponding to the obstacle.

[0088] 2. Tactical Pathfinding and Planning Algorithm

[0089] On the tactical cost map layer established by the above method, the A* path planning algorithm is run to achieve the robot's tactical pathfinding effect. In the tactical pathfinding A* planning algorithm, nodes... n The cost f (n ) Formula as follows:

[0090] f ( n ) = g ( n ) + h ( n )(4)

[0091] wherein, g ( n ) is the movement cost of the robot from the starting point to the node n position, the calculation method in the tactical pathfinding A* algorithm is:

[0092] (5)

[0093] wherein, d k-1,k is the movement distance cost of the node k - 1 to the node k . In addition, the accumulated value of the enemy firepower cost on the path is added on the basis of the original A* algorithm , and the weight parameter K g is used to adjust the weight size of the enemy firepower threat in the navigation planning calculation.

[0094] The heuristic cost function g ( n ) selects the Manhattan distance from the node n to the target point as the prediction approximation to reduce the planning algorithm calculation overhead, and the calculation formula is as follows:

[0095] (6)

[0096] wherein ( x n , y n ), ( x t , y t ) are the coordinates of the node n and the target point t in the map grid.

[0097] 3、Autonomous tactical movement planning

[0098] Beyond basic navigation and movement, combat robots also need to execute specific tactical maneuvers based on mission requirements during autonomous combat. These actions may include approaching and attacking designated enemy units, or retreating to a safe location for self-preservation. The following section will introduce the algorithms for close-range attacks and self-preservation retreats of combat robots, based on tactical cost maps and tactical navigation planning algorithms.

[0099] The close-range attack planning specifies the enemy unit to be attacked, or provides its location coordinates. The robot will autonomously calculate an attack position that can be used to shoot at the enemy unit with an airborne direct-fire weapon as the navigation planning target point, and plan a movement path to reach the target point through the above navigation algorithm.

[0100] The attack location is calculated as follows, based on the enemy unit's coordinates ( x 0 , y Centered on 0), with a side length of 2 d Within the square area, the attack location coordinates are calculated using the following cost formula ( x, y The benefit value of the target is calculated, and the coordinates with the highest benefit value are selected as the attack position.

[0101] (7)

[0102] in, α ( x, y ) is the unobstructed judgment function, when the attack position coordinates ( x, y ) and enemy unit coordinates ( x 0 , y The value is 1 when there is no obstruction between 0 and 1, otherwise the value is 0.

[0103] R ( x, y The gain value for the attack position is determined by the distance between the attack position and the enemy unit. and ideal attack distance The decision is made. The closer the attack initiation distance is to the ideal distance, the higher the benefit. A function is used to calculate the benefit value. for:

[0104] (8)

[0105] K q ·Q ( x, y ) represents the movement cost for the robot to reach the attack location, and represents the robot's current position. x r , y r ) to the attack position ( x 0 , y0) geometric distance, which is an approximation of the actual moving distance. Parameter Kq is used to adjust the weight of the moving cost in the optimization calculation.

[0106] K f · J (x,y) is the threat cost of the attack position to the enemy fire, parameter K f is used to adjust the weight of the fire threat cost in the optimization calculation.

[0107] By continuously performing the above planning algorithm, the combat robot will autonomously approach the vicinity of the attack target, find an attack position that can directly shoot at the enemy unit while reducing the threat of enemy fire, and attack the specified unit.

[0108] When the robot's own condition is in danger, the robot can adopt a retreat and self-protection movement strategy, autonomously move to a safer position according to the position distribution and fire threat distribution of the enemy, and ensure its own safety. The optimization formula for finding the retreat and self-protection movement position is as follows, and the ideal retreat position can be obtained by optimizing the retreat position cost x, y ):

[0109] (9)

[0110] Q ( x, y ) and J ( x, y ) are the moving cost of the robot to reach the retreat point and the enemy fire threat value of the point respectively. Their definition is the same as above.

[0111] By continuously performing the above planning algorithm, the combat robot will autonomously retreat to a safe area, such as behind a shelter or a position away from the enemy, to improve its survival probability.

[0112] In another aspect, the present application also discloses a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to make the processor execute the steps of any of the above methods.

[0113] In another aspect, the present application also discloses a computer device, which comprises a memory and a processor, and the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of any of the above methods.

[0114] In a further embodiment provided in the present application, a computer program product containing instructions, which, when executed on a computer, cause the computer to perform the steps of any of the methods described in the above embodiments, is also provided.

[0115] It can be understood that the system provided in the embodiments of the present application corresponds to the method provided in the embodiments of the present application, and the related content explanation, examples and beneficial effects can refer to the corresponding part in the above method.

[0116] It can be understood by those skilled in the art that all or part of the processes in the above embodiments can be completed by a computer program instructing related hardware, and the program can be stored in a non-volatile computer readable storage medium. When the program is executed, it can include the processes of the above embodiments of the method. Any reference to memory, storage, database or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM) and memory bus dynamic RAM (RDRAM) and the like.

[0117] The technical features of the above embodiments can be combined in any way. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present application.

[0118] The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A robot autonomous navigation algorithm for adversarial environments, characterized in that, Includes the following steps, A navigation tactical cost map containing enemy fire threat information is constructed based on a terrain occlusion determination algorithm. The optimal navigation path is then calculated based on the navigation tactical cost map, and finally, tactical navigation actions that can maximize the safety of oneself are executed. Construct a dynamic tactical action planning algorithm based on tactical benefit analysis to realize close-range attack actions against designated enemy targets, as well as retreat and self-defense commands; The dynamic tactical action planning algorithm also includes autonomous tactical movement planning, specifically close-range attack planning: Close-range attack planning involves specifying the enemy unit to be attacked, or providing its location coordinates. The robot will autonomously calculate an attack position that can be used to fire on the enemy unit with an onboard direct-fire weapon as the navigation planning target point, and plan a movement path to reach the target point through a dynamic tactical action planning algorithm. The attack location is calculated as follows, based on the enemy unit's coordinates ( x 0 , y Centered on 0), with a side length of 2 d Within the square area, the attack location coordinates are calculated using the following cost formula ( x, y The effectiveness value of the target is calculated, and the coordinates with the highest effectiveness value are optimized as the attack position. （7） in, α ( x, y ) is the unobstructed judgment function, when the attack position coordinates ( x, y ) and enemy unit coordinates ( x 0 , y The value is 1 when there is no obstruction between 0 and 1; otherwise, the value is 0. R ( x, y The gain value for the attack position is determined by the distance between the attack position and the enemy unit. and ideal attack distance The closer the attack initiation distance is to the ideal distance, the higher the benefit; the function for calculating the benefit value. for: （8） K q ·Q ( x, y ) represents the movement cost for the robot to reach the attack location, and represents the robot's current position. x r , y r To the attack position ( x 0 , y The geometric distance of 0), which serves as an approximation of the actual distance traveled; parameter K q Used to adjust the weight of the movement cost in the optimization calculation; K f · J (x,y) The cost of enemy fire threat to the attack position, parameter K f Used to adjust the weight of the firepower threat cost in the optimization calculation.

2. The robot autonomous navigation algorithm in an adversarial environment according to claim 1, characterized in that: The tactical cost map is a two-dimensional grid map of the same size as the current map, and the value on each grid represents the cost for the corresponding robot to pass through the location; The tactical cost map is composed of several cost layers stacked together. Each cost layer represents different movement cost information. By stacking and adding all the cost layers together, a complete tactical cost map can be obtained.

3. The robot autonomous navigation algorithm in an adversarial environment according to claim 1, characterized in that: The tactical cost map consists of four cost layers stacked together: an enemy firepower threat layer, a static map layer, an obstacle layer, and an expansion layer. Enemy fire threat layer: Used to represent the spatial distribution of enemy fire threat; For each enemy unit, for all grid points on the layer that are unobstructed from that unit, increase the corresponding fire threat value according to the enemy firepower weight and the inverse square law of distance. Specifically, when the number of enemies is K, for an m × n grid cost layer, the firepower cost J of grid point (i, j) is increased. ij The calculation formula is as follows: （1） （2） in, It is the first k One enemy at the grid point ( i, j The numerical value of the firepower cost caused. Q k For the first k The firepower weight corresponding to each enemy is directly proportional to the firepower intensity of that enemy; For the first One enemy to the grid point The distance; Static map layer: used to represent the movement cost caused by known obstacles on the ground for navigation; The previously input 2D obstacle map of the current environment can be converted into a cost layer. If a grid point is occupied by an obstacle, the cost of that point is set to the preset collision cost; otherwise, the cost is zero. Dynamic obstacle layer: used to represent the movement cost to navigation caused by obstacles scanned by the LiDAR; The obstacle data points scanned by the lidar will be stored in a queue, and the cost of the obstacle data points will be set as the collision cost. Meanwhile, the queue is periodically refreshed to remove old scan data to ensure that the cost information of dynamic obstacles is real-time; Inflated layer: Used to indicate the risk of movement caused by being too close to an obstacle; The inflation cost layer expands the collision cost portion of the static map layer and dynamic obstacle layer by a certain distance in all directions and adds a cost attenuation area. The specific formula for implementing the inflation layer is as follows: (3) Among them, J ij For the cost of the dilation layer at grid point (i, j), d ij It is the distance from grid point (i, j) to the nearest obstacle, r is the set expansion radius, and J0 is the cost value corresponding to the obstacle.

4. The robot autonomous navigation algorithm in an adversarial environment according to claim 3, characterized in that, The dynamic tactical action planning algorithm includes: On the established tactical cost map layer, run the A* path planning algorithm to achieve tactical pathfinding for the robot; in the tactical pathfinding A* planning algorithm, nodes... n The cost f ( n The formula is as follows: f ( n ) = g ( n ) + h ( n )(4) in, g ( n () is the robot moving from the starting point to the node. n The cost of moving a position is calculated in the tactical pathfinding A* algorithm as follows: （5） in, d k-1,k It is a node k - 1 to node k In addition to the movement distance cost, the algorithm also incorporates the accumulated cost of enemy firepower along the path, building upon the original A* algorithm. and using weight parameters K g To adjust the weight of enemy firepower threat in navigation planning calculations; Heuristic cost function g ( n Select nodes n The Manhattan distance to the target point is used as a prediction approximation to reduce the computational cost of the planning algorithm. Its calculation formula is as follows: （6） in( x n , y n ), ( x t , y t ) is a node n and target point t Coordinates in the map grid.

5. The robot autonomous navigation algorithm in an adversarial environment according to claim 1, characterized in that, The autonomous tactical movement planning also includes retreat and self-preservation planning. When the robot is in danger, it can adopt a retreat and self-preservation movement strategy, autonomously moving to a safer location based on the enemy's position distribution and fire threat distribution. The optimization formula for finding a retreat and self-preservation movement location is as follows; optimizing the retreat location cost yields the ideal retreat location. x, y ): （9） Q ( x, y ) and J ( x,y These represent the movement cost for the robot to reach the retreat point and the enemy firepower threat value at that point, respectively.

6. A computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the method as claimed in any one of claims 1 to 5.

Images