Fast Search Path Planning Method and Device in Search and Rescue Environment

Abstract

The present invention provides a method and device for rapid search path planning in a search and rescue environment, including: determining a search and rescue area, dividing the search and rescue area into a plurality of grids, constructing POMDP seven-tuple information corresponding to each grid, where the POMDP seven-tuple information includes a finite state set, a finite action set, a state transition matrix, a reward, a finite observation set, an observation probability, and a discount factor; obtaining the current position information of the search and rescue robot and the surrounding map environment information of the search and rescue robot in real time, determining the grid where the search and rescue robot is currently located based on the current position information, and determining the next target grid of the search and rescue robot through a partially observable Markov decision algorithm based on the POMDP seven-tuple information corresponding to the currently located grid; determining the search path of the search and rescue robot based on the target grid and the surrounding map environment information. This application enables the robot to complete the search work of full-autonomous exploration and mapping while building a map in an unknown environment map, improving the adaptability of the robot to the search and rescue environment.

Description

Technical Field

[0001] The present invention relates to the technical field of robots, and particularly to a method and device for rapid search path planning in a search and rescue environment. Background Art

[0002] With the development of the fields of artificial intelligence and robots, the application fields of unmanned vehicles and unmanned aerial vehicles are becoming more and more extensive, and they have begun to replace people to complete some complex tasks in harsh environments. Using unmanned vehicles, unmanned aerial vehicles, etc. for search and rescue after disasters has been a hot topic of research at home and abroad in recent years.

[0003] An unmanned vehicle is also called a robot. When a robot executes a search and rescue task, it needs to perform path planning. Traditional path planning algorithms include search-based (such as A, A*) methods and sampling-based (RRT, PRM) methods, which can plan a collision-free path from the starting point to the ending point. However, such algorithms cannot fully consider environmental information. In order to enable robots to better execute complex tasks, they must be made more intelligent and more capable of imitating human behavior. Reinforcement learning is a machine learning method. In reinforcement learning, an agent selects actions according to the state of the environment and obtains rewards or punishments from the environment, and improves its decision-making strategy through continuous trial-and-error learning to obtain better long-term rewards. By modeling states, rewards, actions, etc., a robot can be trained to execute complex tasks. However, this method also has great disadvantages. For example, it needs to learn through the interaction between the agent and the environment and requires a large amount of interaction data, which is very expensive and time-consuming in some tasks.

[0004] Although existing robots can achieve path planning in search and rescue tasks, the existing path planning algorithms require the robot to work under the condition of a known environmental map or familiarity with the working environment, which limits its working ability in an unknown and unfamiliar environment. Therefore, in search and rescue tasks, existing robots cannot complete the search and rescue work in an unknown environment. Therefore, how to enable a robot to complete the search work of full-autonomous exploration and mapping while exploring in an unknown environment and improve the adaptability of the robot in the search and rescue environment is a technical problem to be solved urgently. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a method and device for rapid search path planning in a search and rescue environment to eliminate or improve one or more defects existing in the prior art.

[0006] One aspect of the present invention provides a method for rapid search path planning in a search and rescue environment, the method including:

[0007] Determine the search and rescue area, divide the search and rescue area into multiple grids, and construct the POMDP seven-tuple information corresponding to each grid. The POMDP seven-tuple information includes a finite state set, a finite action set, a state transition matrix, a reward, a finite observation set, an observation probability, and a discount factor;

[0008] Obtain the current position information of the search and rescue robot and the surrounding map environment information of the search and rescue robot in real time. Determine the grid where the search and rescue robot is currently located based on the current position information of the search and rescue robot, and determine the next target grid of the search and rescue robot through the partially observable Markov decision algorithm based on the POMDP seven-tuple information corresponding to the currently located grid;

[0009] Determine the search path of the search and rescue robot based on the next target grid of the search and rescue robot and the surrounding map environment information of the search and rescue robot.

[0010] In some embodiments of the present invention, constructing the POMDP seven-tuple information corresponding to each grid includes:

[0011] Determine the search and rescue priority corresponding to each grid, and determine the state set based on the search and rescue priority corresponding to the grid;

[0012] Determine the adjacent grids of the grid, and determine the finite action set based on the adjacent grids;

[0013] Determine the finite observation set corresponding to the grid based on whether there are trapped persons in the grid.

[0014] In some embodiments of the present invention, determining the next target grid of the search and rescue robot through the partially observable Markov decision algorithm based on the POMDP seven-tuple information corresponding to the currently located grid includes:

[0015] Determine the passable area based on the obtained surrounding map environment information of the search and rescue robot;

[0016] Determine the next target grid of the search and rescue robot through the partially observable Markov decision algorithm based on the POMDP seven-tuple information corresponding to the currently located grid and the passable area.

[0017] In some embodiments of the present invention, determining the passable area based on the obtained surrounding map environment information of the search and rescue robot includes:

[0018] Obtain the center points of each grid, and use each center point as each boundary point;

[0019] Determine the connection relationship between the boundary points based on the surrounding map environment information, and the connection relationship is connected, unconnected or unknown;

[0020] Determine the passable area based on the connection relationship between the boundary points.

[0021] In some embodiments of the present invention, the current position information of the search and rescue robot and the surrounding map environment information of the search and rescue robot are obtained in real time, including:

[0022] Based on the Simultaneous Localization and Mapping (SLAM) method, the current position information of the search and rescue robot and the surrounding map environment information of the search and rescue robot are obtained in real time.

[0023] In some embodiments of the present invention, based on the target grid of the next step of the search and rescue robot and the surrounding map environment information of the search and rescue robot, the search path of the search and rescue robot is determined, including:

[0024] Based on the target grid of the next step of the search and rescue robot and the surrounding map environment information of the search and rescue robot, the search path of the search and rescue robot is determined by the robot autonomous navigation algorithm.

[0025] In some embodiments of the present invention, there are multiple search and rescue robots, and multiple search and rescue robots communicate with a common information processing node, and the current position information, the surrounding map environment information, and the search path are shared among the multiple search and rescue robots.

[0026] In some embodiments of the present invention, the current position information, the surrounding map environment information, and the search path are shared among the multiple search and rescue robots, including:

[0027] Obtain the current position information, the surrounding map environment information, and the search path corresponding to each search and rescue robot;

[0028] Fuse the current position information, the surrounding map environment information, and the search path corresponding to each search and rescue robot based on the common information processing node to obtain fused information;

[0029] Send the fused information to each search and rescue robot.

[0030] According to another aspect of the present invention, a fast search path planning system in a search and rescue environment is also disclosed. The system includes a processor, a memory, and a computer program stored on the memory. The processor is used to execute the computer program, and when the computer program is executed, the system implements the steps of the method described in any of the above embodiments.

[0031] According to still another aspect of the present invention, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in any of the above embodiments are implemented.

[0032] The rapid search path planning method and device in the search and rescue environment disclosed in the above embodiments of the present invention first divide the search and rescue environment into grids, and then construct the POMDP seven-tuple information for each grid, so as to calculate the action with the maximum reward for the next step of the search and rescue robot based on the partially observable Markov decision algorithm; and during the search and rescue process, the method of building a map while exploring is adopted to complete the planning of the search path. This method can realize the fully autonomous search operation of exploring and building a map while the environment is partially known and the map is unknown, and improves the adaptability of the robot in the search and rescue environment.

[0033] In addition, the rapid search path planning method in the present application can use multiple search and rescue robots to work together, and the multiple search and rescue robots share the current position information, the surrounding map environment information, and the search path, which further improves the search and rescue efficiency and reduces the search and rescue time.

[0034] The additional advantages, objects, and features of the present invention will be partially described below, and will become partially apparent to those of ordinary skill in the art after studying the following text, or can be learned according to the practice of the present invention. The objects and other advantages of the present invention can be achieved and obtained by the structure specifically pointed out in the specification and the drawings.

[0035] Those skilled in the art will understand that the objects and advantages that can be achieved by the present invention are not limited to the above specifically described, and the above and other objects that the present invention can achieve will be more clearly understood according to the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings described herein are used to provide a further understanding of the present invention, and constitute a part of this application, and do not limit the present invention. The components in the drawings are not drawn to scale, but are only for showing the principles of the present invention. In order to facilitate showing and describing some parts of the present invention, the corresponding parts in the drawings may be enlarged, that is, may become larger relative to other components in the exemplary device actually manufactured according to the present invention. In the drawings:

[0037] Figure 1 It is a schematic flow chart of the rapid search path planning method in the search and rescue environment according to an embodiment of the present application.

[0038] Figure 2a It is a schematic diagram of the search and rescue area of the rapid search path planning method according to an embodiment of the present application.

[0039] Figure 2b It is a schematic diagram of the grid map of the rapid search path planning method according to an embodiment of the present application.

[0040] Figure 3a It is a schematic diagram of the moving direction of the search and rescue robot according to an embodiment of the present applicationFigure 1 .

[0041] Figure 3b Schematic diagram II of the moving direction of the search and rescue robot according to an embodiment of the present application.

[0042] Figure 4a Map schematic used during the simulation experiment Figure 1 .

[0043] Figure 4b Map schematic diagram II used during the simulation experiment

[0044] Figure 5a Schematic diagram of the search and rescue process of the search and rescue robot during the simulation experiment Figure 1 .

[0045] Figure 5b Schematic diagram II of the search and rescue process of the search and rescue robot during the simulation experiment

[0046] Figure 5c Schematic diagram III of the search and rescue process of the search and rescue robot during the simulation experiment

[0047] Figure 5d Schematic diagram IV of the search and rescue process of the search and rescue robot during the simulation experiment

[0048] Figure 6 Schematic diagram of the communication structure of multiple search and rescue robots according to an embodiment of the present application Detailed implementation manners

[0049] To make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the implementation manners and the drawings. Herein, the illustrative implementation manners of the present invention and their descriptions are used to explain the present invention, but do not limit the present invention.

[0050] Herein, it should also be noted that in order to avoid obscuring the present invention due to unnecessary details, only the structures and / or processing steps closely related to the solution according to the present invention are shown in the drawings, while other details less related to the present invention are omitted.

[0051] It should be emphasized that the term "comprising / including" when used herein refers to the presence of features, elements, steps or components, but does not exclude the presence or addition of one or more other features, elements, steps or components.

[0052] Herein, it should also be noted that if not otherwise specified, the term "connection" in this document can not only refer to a direct connection, but also represent an indirect connection with an intermediate, and can not only represent a wired connection, but also a wireless connection, and can be specifically changed based on the actual application scenario.

[0053] Simulate a real search and rescue scenario. When using a search and rescue robot for search and rescue, every location in the scenario needs to be searched in the shortest time to find all the trapped people. This process can be modeled as full coverage path planning. The full coverage path planning algorithm is a path planning algorithm used in automated systems, aiming to ensure that the search and rescue robot can cover all points or areas in a given environment. Its goal is to find a path that enables the search and rescue robot to pass through each point or area on this path, thereby achieving full coverage of the entire environment.

[0054] The fast search path planning method in the search and rescue environment of this application applies decision-making planning to the search and rescue scenario. At the same time, it combines an autonomous detection algorithm to enable the robot to conduct inspections on every location in the environment when there is no map and the environmental situation is partially known. This application can enable the robot to achieve fully autonomous search operations of exploring while building a map in an unknown environment, pass through all grid nodes on the path in the shortest possible time, and can improve the search and rescue efficiency by means of the collaborative work of multiple search and rescue robots.

[0055] To better understand the present invention, the following explanations are made for related terms:

[0056] Partially Observable Markov Decision Process (POMDP) is an extension of Markov Decision Process (MDP) used to solve decision-making problems in partially observable environmental states. By modeling its seven-tuple <S, A, T, R, Ω, O, λ>, it can well simulate the uncertainty in the environment. In POMDP, the environment faced by the decision maker is random and cannot be fully observed, and can only be inferred through a series of observation variables; at each moment, the system generates an observation variable based on the current state and the executed action. The observation variable provides partial information about the system state. Then the decision maker updates its belief state according to the observation variable. The belief state is the probability value of each state, and then POMDP is solved to select the next action. The goal of the decision maker is to find an optimal strategy to maximize the long-term cumulative reward.

[0057] In the following text, embodiments of the present invention will be described with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0058] Figure 1 It is a flowchart of the fast search path planning method in the search and rescue environment according to an embodiment of this application. As shown in the figure, the fast search path planning method includes at least steps S10 to S30.

[0059] Step S10: Determine the search and rescue area, divide the search and rescue area into multiple grids, and construct the POMDP seven-tuple information corresponding to each grid. The POMDP seven-tuple information includes a finite state set, a finite action set, a state transition matrix, a reward, a finite observation set, an observation probability, and a discount factor.

[0060] In this step, the search and rescue environment is divided into grids to generate a grid map. Each grid represents an area in the environment and serves as a state of the POMDP, that is, the POMDP is used to model each grid. The modeling information includes possible actions of the robot, observation results, state transition probabilities, etc. Figure 2a The figure shows a schematic diagram of the search and rescue area. This search and rescue map can be established based on Gazebo. The map size can be 20m * 20m, and obstacles are randomly placed in the map; further, this search and rescue map can be converted into a 10 * 10 grid map, with each grid size being 2m * 2m. The corresponding schematic diagram of the grid map is as Figure 2b shown. In Figure 2b it, the solid lines in the two grids are obstacles. The solid lines indicate impassable, and the dashed lines indicate passable. In this step, the grid map is modeled according to the POMDP seven-tuple to simulate the real search and rescue environment, so that the robot continuously plans paths at the starting point to traverse all grids in the shortest possible time.

[0061] In a real search and rescue scenario, the probability of trapped people existing in each place is different. For example, during an earthquake, the probability of people existing under a collapsed building is higher than in the wild. Therefore, the priority of a collapsed building is higher during search and rescue. This application assigns a priority to each grid in the grid map based on the actual search and rescue scenario, representing its importance. The higher the reward obtained when passing through a grid with a higher priority. In addition, the probability of observing things in grids with different priorities is also different, and an observation variable will be assigned to each grid. The observation variable is an element in the observation space set.

[0062] During the process of POMDP seven-tuple modeling, the probability distribution of the positions of the people to be rescued is considered. Dividing the grids into different priorities represents the urgency of the grids. The higher the priority, the greater the probability of trapped people existing, and the greater the set reward. Two observation information (o1, o2) are set, and different observation functions are assigned; the observation information is set to different values according to specific circumstances. o1 is more associated with high priorities, that is, the probability of observing o1 in grids with higher priorities is greater, and the probability of observing 02 in grids with lower priorities is greater.

[0063] Exemplarily, constructing the POMDP seven-tuple information corresponding to each grid includes: determining the search and rescue priority corresponding to each grid, and determining the state set based on the search and rescue priority corresponding to the grid; determining the adjacent grids of the grid, and determining the finite action set based on the adjacent grids; determining the finite observation set corresponding to the grid based on whether there are trapped persons in the grid. In this embodiment, the search and rescue priority corresponding to each grid can be determined based on the terrain distribution of the search and rescue area; exemplarily, the search and rescue priority can be divided into levels S1, S2, and S3. The search and rescue priority is the emergency level of the search and rescue site. The higher the search and rescue priority, the greater the probability of there being trapped persons, the more urgent the terrain, and the more it should be searched first; moreover, the distribution of terrains with different priorities is also different, and the specific priority of each grid can be determined according to the actual situation.

[0064] Specifically, in the POMDP seven-tuple <S, A, T, R, Ω, O, λ>, the finite state set S = <S1, S2, S3> represents the probability that the grid belongs to each priority level; the finite action set A = <a1, a2, a3, a4, a5, a6, a7, a8> represents the 8 neighboring grids around a certain grid. When the robot moves to a certain grid, it can move to the next grid in 8 directions around this grid in the next step. The state transition matrix represents the probability of reaching the next state by taking action a in state s, that is, the probability of moving from one priority level to another through a certain direction of movement. This probability is related to the priority assignment of the environment. The probability of moving from one priority level to another through a certain direction of movement is related to the priority assignment of the environment. represents the reward for reaching state s by taking action a; since the priorities of S1, S2, and S3 decrease gradually, the rewards for moving to S1, S2, and S3 also decrease gradually. The finite observation set Ω = <o1, o2> represents the information on whether there are trapped persons in the observed grid, that is, the set of phenomena observed by the robot during movement. It can be the picture taken by the camera or the data collected by the sensor. When the robot moves to a certain grid, it will obtain an observation variable, and the observation variable obtained when passing through a certain grid repeatedly remains unchanged, that is, the data collected by the sensor at the same location does not change. The observation function O = pr(o|s, a) represents the probability of observing the observation variable o when taking action a to reach state s. The distribution of the finite observation set is related to the observation function and the settings of the environment.

[0065] Step S20: Real-time obtain the current position information of the search and rescue robot and the surrounding map environment information of the search and rescue robot, determine the grid where the search and rescue robot is currently located based on the current position information of the search and rescue robot, and determine the target grid for the next step of the search and rescue robot through the partially observable Markov decision algorithm based on the POMDP seven-tuple information corresponding to the currently located grid.

[0066] In this step, based on the current position and status of the robot and the map information constructed in real time by the lidar, and using the POMDP model to calculate the action with the maximum reward for the next step; after executing this action, update the status of the robot according to the obtained environmental information to achieve rapid traversal and search of every corner in the environment. In this step, based on the solution of the partially observable Markov decision algorithm (POMDP algorithm), determine the decision of the robot at each step in the unknown environment.

[0067] In some embodiments of the present invention, determining the target grid of the search and rescue robot for the next step based on the POMDP seven-tuple information corresponding to the currently located grid through the partially observable Markov decision algorithm includes: determining the passable area based on the obtained surrounding map environment information of the search and rescue robot; determining the target grid of the search and rescue robot for the next step through the partially observable Markov decision algorithm based on the POMDP seven-tuple information corresponding to the currently located grid and the passable area.

[0068] Further, determining the passable area based on the obtained surrounding map environment information of the search and rescue robot includes the following steps: obtaining the center points of each grid and taking each of the center points as a boundary point; determining the connection relationship between the boundary points based on the surrounding map environment information, where the connection relationship is connected, unconnected, or unknown; determining the passable area based on the connection relationship between the boundary points.

[0069] When the search and rescue robot performs tasks in an unknown environment map, it first needs to obtain map and other information of the environment as soon as possible. However, in the prior art, the main way to construct a map is to manually control the movement of the robot, that is, traverse all unknown environments, collect environmental feature information through sensors, and then construct a global map based on the environmental feature information collected by the sensors. This method requires manual control throughout the process and is impossible to implement in a search and rescue scenario. In order to further improve the intelligence of the search and rescue robot, this application enables the search and rescue robot to obtain the most complete and accurate environmental map in a limited time and without human intervention based on the robot's autonomous exploration technology. Many existing map exploration strategies are based on boundaries, where the boundary is defined as the dividing line between the unknown space and the known space; the idea of the boundary-based exploration strategy is to guide the robot to the unknown area to complete the exploration task. Therefore, the autonomous exploration task is generally divided into three steps: generating boundary points, selecting the boundary point with the highest evaluation value, and planning a path to the selected boundary point. The selection of the target boundary point is the key to effective exploration. The boundary-based exploration algorithm detects the boundary points between the explored area and the unknown area, processes and filters the boundary points, and takes the boundary point with the maximum comprehensive benefit as the target point for continuous exploration until there are no boundary points to complete the exploration.

[0070] In the above embodiments, the fast search path planning method obtains the current position information of the search and rescue robot and the surrounding map environment information of the search and rescue robot in real time based on the simultaneous localization and mapping method. Further, the passable area is determined based on the obtained surrounding map environment information of the search and rescue robot. The selection of boundary points needs to comprehensively consider information gain, navigation cost, and robot positioning accuracy, and this process consumes certain resources; when determining the passable area in this application, the center point of each grid in the grid map can be first selected as the boundary point, and the boundary points of each map are determined. In the grid map shown in Figure 2b there are 100 grids, so there are 100 boundary points; the connection relationship between boundary points is connected (there is no obstacle between two boundary points), unconnected (there is an obstacle between two boundary points), or unknown (not scanned by the radar); during the exploration process, the search and rescue robot continuously scans whether it is connected between all boundary points through the lidar to construct a passable area, and the robot can plan a path in this passable area and execute tasks. At the beginning, the passable area is very small. As the robot moves continuously, the range scanned by the radar becomes larger and larger, and the passable area gradually becomes larger.

[0071] During the search process of the search and rescue robot in the above embodiments, environmental information is obtained through sensors such as radar or cameras, and an environmental map is constructed in real time; this step uses SLAM technology to achieve simultaneous localization and mapping of the robot in an unknown environment. In addition, this application can use the open-source framework gmapping for environmental modeling, build a map during the continuous movement of the robot, and construct a passable area. The passable area is as shown in Figure 3a and Figure 3b ; in Figure 3a , the gray lines connect all known areas together to form a passable area, and path navigation of the search and rescue robot is realized based on this passable area, so that the robot can move freely in the passable area, and each decision will move to the next open grid. During the process of the robot continuously executing tasks and building a map, the passable area will continue to increase, as shown in Figure 5a , Figure 5b , Figure 5c , Figure 5d , shown in Figures 5a to 5d , the passable area gradually increases. It can be seen from this that based on the above grid scanning algorithm, it can be ensured that the robot completes navigation by using the map information constructed in real time in each step, that is, it ensures that the search and rescue robot can explore while building a map without an environmental map, and each decision makes full use of the known part of the environment and map information, and at the same time ensures that the overall fast search path planning is independently completed by the search and rescue robot without relying on manual navigation.

[0072] Step S30: Determine the search path of the search and rescue robot based on the next target grid of the search and rescue robot and the surrounding map environment information of the search and rescue robot.

[0073] In this step, a specific search path is determined. This search path enables the robot to walk towards the specified target point without colliding with obstacles. When the search and rescue robot performs a search and rescue mission, it is necessary to ensure that the grid with a higher priority is searched first and all grids are searched as soon as possible. In an embodiment, the search path of the search and rescue robot is determined based on the next target grid of the search and rescue robot and the surrounding map environment information of the search and rescue robot through the robot's autonomous navigation algorithm. The purpose of the autonomous navigation algorithm is to enable the robot to effectively explore and navigate in an unknown environment.

[0074] Exemplarily, after obtaining the optimal action of the search and rescue robot, the navigation of the search and rescue robot can be further realized through the ROS system, specifically using function packages such as amcl and move_base in the ROS system. The amcl function package will perform real-time positioning of the robot based on the sensor data carried by the robot (such as lidar scan data), while the move_base function package is responsible for planning a collision-free path from the current position to the target position and publishing the cmd_vel velocity control topic to drive the movement of the robot.

[0075] In some embodiments of the present invention, there are multiple search and rescue robots. All of the multiple search and rescue robots communicate with a common information processing node, and the current position information, surrounding map environment information, and search path are shared among the multiple search and rescue robots. In this embodiment, through the collaborative work of multiple search and rescue robots, the search and rescue efficiency is further improved and the search and rescue time is reduced. Moreover, all of the multiple search and rescue robots use the fast search path planning method disclosed in any of the above embodiments to implement path planning.

[0076] Furthermore, the sharing of the current position information, surrounding map environment information, and search path among the multiple search and rescue robots may specifically include: obtaining the current position information, surrounding map environment information, and search path corresponding to each search and rescue robot; fusing the current position information, surrounding map environment information, and search path corresponding to each search and rescue robot based on the common information processing node to obtain fused information; and sending the fused information to each search and rescue robot.

[0077] In addition, to further improve the search and rescue efficiency of multiple search and rescue robots, it is necessary to improve the communication efficiency between them. In this application, each search and rescue robot needs to know the information of other robots during the mission execution. If connections are to be established between every two robots, the entire communication system will be very complex and inflexible when adding or removing robots. Therefore, this application adopts a communication method similar to the bus bus to realize the communication of multiple search and rescue robots. Robots exchange information by publishing topics or service communications. The schematic diagram of the communication structure is shown in Figure 5. At this time, each robot publishes information such as its own mapping and exploration results to the public information processing node. After the public information processing node fuses this information, it sends the fused information to each robot in the form of topic or service communication, so that each robot has the vision of all robots. Based on this method, we can flexibly add any number of search and rescue robots.

[0078] As can be seen from the above embodiments, the fast search path planning method of this application uses the partially observable Markov decision method to determine the decision of the robot at each step in the unknown environment, and uses the simultaneous localization and mapping technology for mapping, that is, an algorithm for the robot to explore and map simultaneously and autonomously in the unknown map environment. According to partial information of the environment, it quickly searches every place in the environment to determine whether there are people in need of rescue.

[0079] To verify the effectiveness of the fast search path planning method of this application, this application further uses the Robot Operating System (ROS) as a software framework, models the robot and the search and rescue environment in Gazebo and conducts simulation experiments, and observes the experimental results through the rviz plugin. The schematic diagram of the map used in the simulation experiment is as Figure 4a and Figure 4b shown. Based on the simulation results, it is found that the fast search path planning method of this application can enable the robot to complete the search work of exploring and mapping simultaneously and autonomously in the unknown environment, improving the adaptability of the robot in the search and rescue environment.

[0080] Correspondingly, the present invention also provides a fast search path planning system in a search and rescue environment. The system includes a processor, a memory, and a computer program stored on the memory. The processor is used to execute the computer program, and when the computer program is executed, the system implements the steps of the method described in any of the above embodiments.

[0081] Exemplarily, the fast search path planning system may specifically further include a path planning module, a robot navigation module, a simultaneous localization and mapping (SLAM) module, a multi-robot collaboration module, and a simulation experiment module. The functions of each module are as follows:

[0082] Path Planning Module: Divide the map to generate a grid map, and use the Partially Observable Markov Decision Process (POMDP) to perform a seven-element modeling for each grid, and finally solve it to determine the decision of the robot at each step in the unknown environment. According to the current position, state of the robot and partial information of the environment, this module calculates the optimal next action to achieve a fast traversal and search of every corner in the environment.

[0083] Robot Navigation Module: Use function packages such as amcl and move_base in the ros system to achieve the navigation of the robot. Amcl can achieve the positioning of the robot, and move_base can plan a collision-free path from the starting point to the ending point and publish a velocity topic to control the movement of the robot.

[0084] Simultaneous Localization and Mapping (SLAM) Module: During the search process of the robot, obtain environmental information through sensors such as radar or camera, and build an environmental map in real time. This module uses SLAM technology to achieve simultaneous localization and mapping of the robot in the unknown environment. This application uses the open-source framework gmapping for environmental modeling, and the grid scanning algorithm proposed based on the above content ensures that the robot always uses the map information built in real time.

[0085] Multi-Robot Collaboration Module: Add multiple robots to improve the search and rescue efficiency. When multiple robots execute tasks, they use the bus communication method. The robots publish their own information to the information processing node using topics or service communication. After the information processing node fuses and processes the information, it sends it to each robot using topics or service communication to ensure effective information exchange and collaborative work among the robots. Moreover, using this communication method, any number of robots can be added or cancelled flexibly.

[0086] Simulation Experiment Module: Use the Robot Operating System (ROS) as the software framework, model the robot and the search and rescue environment in gazebo and conduct simulation experiments, and observe the experimental results through the rviz plugin. This module simulates the search and rescue environment and verifies the effectiveness and performance of the fast search path planning method of this application.

[0087] In addition, the fast search path planning system may also include a user interface and a visualization interface. The user interface is used to accept the search and rescue instructions input by the user, and the visualization interface is used to display the actions and planning results of the robot at each step in real time.

[0088] As can be seen from the above embodiments, for the rapid search path planning method and device in the search and rescue environment of the present application, when performing path planning, a priority map is constructed to simulate the real search and rescue environment as much as possible, the POMDP septuple is modeled, and dynamic and efficient solution is achieved based on the solution method of POMDP. That is, during the process of the robot executing the task, continuous solution is carried out. That is, when the robot moves to a certain grid, an observation variable is observed, its belief state is updated according to the observation function, and then the solution of POMDP is carried out, that is, the action with the maximum long-term reward is selected, and finally it moves to the next grid according to its action. In addition, the present application also realizes the autonomous exploration of the robot when the map is unknown. While building the map, a passable area is generated to ensure that the robot makes decisions within the passable area, realizing the full-autonomous path planning of the robot. This ensures that the robot executes the search and rescue task in the shortest possible time and minimizes the backtracking phenomenon. That is, the robot can not only quickly search the entire map, but also search each grid as few times as possible.

[0089] Specifically, the rapid search path planning method of the present application uses the POMDP decision-making and planning algorithm model for the robot to achieve intelligent search and rescue. The idea of priority in the search and rescue scenario is proposed, that is, the higher the priority, the more it should be searched first. By setting the reward function in the septuple, it can be ensured that the robot first focuses on the grids with high priority, but the influence of distance also needs to be considered. That is, when the robot is far away from the grid with high priority, the travel consumption of moving past will be considered. Therefore, when the robot executes the search and rescue task, when selecting the next grid to move to, it will take into account both the movement consumption and the search and rescue efficiency. At the same time, the POMDP model can well simulate the uncertainties that may occur in the real environment.

[0090] The embodiment of the present invention also provides a computer-readable storage medium and a computer program product, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in any of the above embodiments are implemented. The computer-readable storage medium may be a tangible storage medium, such as a random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, register, floppy disk, hard disk, removable storage disk, CD-ROM, or any other form of storage medium well-known in the technical field.

[0091] Those of ordinary skill in the art should understand that the various exemplary components, systems, and methods described in connection with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Specifically, whether to implement it in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, and so on. When implemented in software, the elements of the present invention are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or a communication link.

[0092] It should be clear that the present invention is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present invention is not limited to the specific steps described and illustrated. Those skilled in the art can make various changes, modifications, and additions, or change the order between steps after understanding the spirit of the present invention.

[0093] In the present invention, the features described and / or illustrated for one embodiment can be used in the same or a similar manner in one or more other embodiments, and / or combined with the features of other embodiments or replace the features of other embodiments.

[0094] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, various changes and modifications can be made to the embodiments of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for rapid search path planning in a search and rescue environment, characterized in that: The method comprises: Determine a search and rescue area, divide the search and rescue area into a plurality of grids, and construct POMDP seven-tuple information corresponding to each grid, wherein the POMDP seven-tuple information includes a finite state set, a finite action set, a state transition matrix, a reward, a finite observation set, an observation probability, and a discount factor; Acquire the current position information of the search and rescue robot and the surrounding map environment information of the search and rescue robot in real time, determine the current grid of the search and rescue robot based on the current position information of the search and rescue robot, and determine the next target grid of the search and rescue robot through a partially observable Markov decision algorithm based on the POMDP seven-tuple information corresponding to the current grid; Determine the search path of the search and rescue robot based on the next target grid of the search and rescue robot and the surrounding map environment information of the search and rescue robot; Construct the POMDP seven-tuple information corresponding to each grid, including: Determine the search and rescue priority corresponding to each of the grids, and determine the state set based on the search and rescue priority corresponding to the grid; Determine adjacent grids of the grid, and determine the limited action set based on the adjacent grids; Determine a finite observation set corresponding to the grid based on whether there is a trapped person in the grid; Among them, the search and rescue priority corresponding to each grid is determined based on the terrain distribution of the search and rescue area, and the limited observation set includes two different observation information o1 and o2. The higher the priority grid, the greater the probability of observing o1, and the lower the priority grid, the greater the probability of observing o2.

2. The method for rapid search path planning in a search and rescue environment according to claim 1, characterized in that: Determining the next target grid of the search and rescue robot through a partially observable Markov decision algorithm based on the POMDP seven-tuple information corresponding to the current grid, including: Determining a traversable area based on the acquired surrounding map environment information of the search and rescue robot; The next target grid of the search and rescue robot is determined by a partially observable Markov decision algorithm based on the POMDP seven-tuple information corresponding to the current grid and the traversable area.

3. The method for rapid search path planning in a search and rescue environment according to claim 2, characterized in that: Determining a traversable area based on the acquired surrounding map environment information of the search and rescue robot includes: Obtain the center point of each grid, and use each center point as each boundary point; Determine a connection relationship between boundary points based on the surrounding map environment information, where the connection relationship is connected, disconnected or unknown; A passable area is determined based on the connection relationship between the boundary points.

4. The method for rapid search path planning in a search and rescue environment according to claim 1, characterized in that: Real-time acquisition of the current location information of the search and rescue robot and the surrounding map environment information of the search and rescue robot includes: Based on the instant positioning and map building method, the current position information of the search and rescue robot and the surrounding map environment information of the search and rescue robot are obtained in real time.

5. The method for rapid search path planning in a search and rescue environment according to claim 1, characterized in that: Determining a search path of the search and rescue robot based on the next target grid of the search and rescue robot and surrounding map environment information of the search and rescue robot includes: The search path of the search and rescue robot is determined by a robot autonomous navigation algorithm based on the target grid of the search and rescue robot's next step and the surrounding map environment information of the search and rescue robot.

6. The method for rapid search path planning in a search and rescue environment according to any one of claims 1 to 5, characterized in that: There are multiple search and rescue robots, and the multiple search and rescue robots all communicate with a common information processing node, and the multiple search and rescue robots share current location information, surrounding map environment information and search paths.

7. The method for rapid search path planning in a search and rescue environment according to claim 6, characterized in that: The multiple search and rescue robots share current location information, surrounding map environment information and search paths, including: Obtaining current location information, surrounding map environment information, and search paths corresponding to each of the search and rescue robots; The current position information, surrounding map environment information and search path corresponding to each of the search and rescue robots are integrated based on the common information processing node to obtain integrated information; The fused information is sent to each of the search and rescue robots.

8. A fast search path planning system in a search and rescue environment, the system comprising a processor, a memory and a computer program stored in the memory, characterized in that: The processor is used to execute the computer program. When the computer program is executed, the system implements the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images