Collaborative navigation method and system for odor tracing drone and search robot

Abstract

The present invention provides a collaborative navigation method and system for an odor tracing drone and a search robot, and relates to the field of intelligent search and rescue technology. The present invention uses information entropy as the main basis for odor navigation, so that the search robot can still operate normally and navigate accurately in environments with poor visibility such as smoke, night or low light, thereby expanding the application scenarios of the search robot.

Description

Technical Field

[0001] The present invention relates to the field of intelligent search and rescue technology, and in particular to a collaborative navigation method and system for an odor tracing drone and a search robot. Background Art

[0002] At present, intelligent search and rescue systems rely on information provided by multiple sensors such as visual sensors, radars, and lidars to make decisions. Although this multi-source information fusion method has improved the system's perception ability and accuracy to a certain extent, it also brings high requirements and high dependence on information sources. Once an information channel fails or data is missing, such as the performance of visual sensors degrades in harsh environments, radar signals are interfered with, etc., the stability and reliability of the entire system will be seriously affected. That is, the existing intelligent search and rescue system has low stability. Summary of the invention

[0003] 1. Technical issues to be solved

[0004] In view of the shortcomings of the prior art, the present invention provides a collaborative navigation method and system for an odor tracing UAV and a search robot, which solves the technical problem of low stability of the existing intelligent search and rescue system.

[0005] (II) Technical solution

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0007] In a first aspect, the present invention provides a collaborative navigation method for an odor tracing drone and a search robot, comprising:

[0008] S1, obtain the real-time environmental information and odor data of the current location collected by the drone, and convert the environmental information into a digital map;

[0009] S2. Using the digital map and the execution parameters of the algorithm as input data, and according to the odor data of the current position, calculating the direction in which the information entropy is reduced the fastest through the information trend algorithm, and using this direction as the flight direction of the UAV during the search process; determining the odor detection point on the digital map during the flight of the UAV in the flight direction;

[0010] S3, determining whether the odor detection point is a qualified point, if so, planning a path between two qualified points through a path planning algorithm to form an initial path for the search robot, and returning to step S2;

[0011] S4. Smoothing the initial path to form a navigation path for the search robot.

[0012] Preferably, S2 specifically includes:

[0013] S201, using the digital map and the execution parameters of the algorithm as input data, and calculating the information entropy distribution through the information trend algorithm according to the odor data of the current location;

[0014] S202, determining the direction in which the information entropy decreases fastest as the flight direction of the UAV during the search process;

[0015] S203, introducing Monte Carlo random sampling to determine the odor detection points on the digital map when the drone is flying in the flight direction.

[0016] Preferably, the S203 includes:

[0017] Introducing Monte Carlo random sampling, if M sampling points are randomly selected According to the Monte Carlo method, the posterior probability distribution P(X 0:k |Y 1:k ) is approximately:

[0018]

[0019] Among them, X k is the sampling point distribution at time k, is the position of the mth particle at time k, Y k is the state observation value at time k, and δ(·) represents the Dirac function; thus, the probability map of the odor source distribution can be constructed;

[0020] After the probability map is constructed, the search robot is set to follow the search route of the drone and start moving in the map from the starting point; when the search robot moves to a certain position r, the position r of the plume source in the gas diffusion model is combined 0 Calculate the expected hit rate R(r|r 0 ), then the expected average number of hits per unit time Δt is:

[0021] k=ΔtR(r|r 0 )

[0022] This constructs a Poisson distribution model:

[0023]

[0024] The hit is simulated by judging whether the Poisson distribution random number z is greater than 0, that is, whether the smell is detected; if the smell is detected, the detection point is recorded, otherwise, the information entropy distribution is recalculated until the search stops at the location of the smell source.

[0025] Preferably, the step of determining whether the odor detection point is a qualified point includes:

[0026] Determine whether the odor detection point overlaps with the position of the obstacle, or whether the distance between the current odor detection point and the previous odor detection point is less than or equal to the threshold; if one of the above conditions is met, it is an unqualified point; if neither condition is met, it is a qualified point.

[0027] Preferably, the path planning algorithm includes a fast traversal random tree algorithm.

[0028] Preferably, the smoothing of the initial path includes: using a path smoothing technique based on a B-spline curve to smooth the initial path.

[0029] Preferably, it also includes:

[0030] S5. Obtain the data after the search robot searches along the navigation path, and determine whether the search robot has reached the odor source. If so, end the search; otherwise, return to step S2.

[0031] In a second aspect, the present invention provides a collaborative navigation system of an odor tracing drone and a search robot, comprising:

[0032] The data acquisition module is used to execute S1, obtain the real-time environmental information collected by the drone and the odor data of the current location, and convert the environmental information into a digital map;

[0033] The odor detection point determination module is used to execute S2, take the digital map and the execution parameters of the algorithm as input data, calculate the direction in which the information entropy is reduced the fastest according to the odor data of the current position through the information trend algorithm, and use it as the flight direction of the UAV during the search process; determine the odor detection point on the digital map during the flight of the UAV according to the flight direction;

[0034] A path planning module is used to execute S3, determine whether the odor detection point is a qualified point, and if so, plan a path between two qualified points through a path planning algorithm to form an initial path of the search robot, and return to step S2;

[0035] The path smoothing module is used to execute S4, smooth the initial path, and form a navigation path for the search robot.

[0036] In a third aspect, the present invention provides a computer-readable storage medium storing a computer program for collaborative navigation between an odor-tracing drone and a search robot, wherein the computer program enables a computer to execute the collaborative navigation method between an odor-tracing drone and a search robot as described above.

[0037] In a fourth aspect, the present invention provides an electronic device, comprising:

[0038] One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the programs include methods for executing the collaborative navigation method of the odor tracing drone and the search robot as described above.

[0039] (III) Beneficial effects

[0040] The present invention provides a collaborative navigation method and system for an odor tracing drone and a search robot. Compared with the prior art, the present invention has the following beneficial effects:

[0041] The present invention adopts information entropy as the main basis for odor navigation, so that the search robot can still operate normally and navigate accurately in environments with poor visibility such as smoke, night or low light, thereby expanding the application scenarios of the search robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0043] Figure 1 A block diagram of a collaborative navigation method between an odor tracing drone and a search robot according to an embodiment of the present invention;

[0044] Figure 2 This is a specific flow chart of a collaborative navigation method of an odor tracing drone and a search robot according to an embodiment of the present invention;

[0045] Figure 3 A schematic diagram of a digital map for visualization;

[0046] Figure 4 A schematic diagram of the search path. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] The embodiments of the present application solve the technical problem of low stability of existing intelligent search and rescue systems by providing a collaborative navigation method and system for an odor tracing UAV and a search robot, realize the use of information entropy as the main basis for odor navigation, improve the stability of the intelligent search and rescue system, and enable the search robot to operate normally and navigate accurately in environments with poor visibility such as smoke, night or low light, thereby expanding the application scenarios of the search robot.

[0049] The technical solution in the embodiment of the present application is to solve the above technical problems, and the overall idea is as follows:

[0050] The collaborative navigation method and system of the odor tracing drone and the search robot in the embodiment of the present invention involve intelligent search and rescue technology and odor tracing technology. The existing intelligent search and rescue technology and odor tracing technology mainly have the following defects:

[0051] 1. Existing technologies generally rely on information provided by multiple sensors such as visual sensors, radars, and laser radar (LiDAR) for decision making. Although this multi-source information fusion method has improved the system's perception ability and accuracy to a certain extent, it also brings high requirements and high dependence on the source of information. Once an information channel fails or data is missing, such as the performance of visual sensors degrades in harsh environments or radar signals are interfered with, the stability and reliability of the entire system will be seriously affected. This high degree of dependence not only increases the complexity and cost of the system, but also limits its application capabilities in scenarios where information acquisition is limited or there is a single information source;

[0052] 2. Existing technologies have mostly adopted behavioral imitation strategies and model inversion strategies for odor tracing. However, in real complex environments, the search strategy based on imitation of biological olfactory behavior has a relatively low success rate in turbulent environments or sparse environments with very low odor concentrations. The strategy based on model inversion is the mathematical and engineering implementation of smoke plume tracking. Not only does the algorithm performance rely heavily on the accuracy of the model's description of the smoke plume environment, but it also needs to be coordinated with the corresponding data collection strategy. At the same time, due to the instability of the turbulent environment, search robots using these two strategies are prone to fall into local mechanism areas and cannot quickly and accurately complete the odor tracing task. The existing technology generally overcomes the above problems by using information trending methods, such as CN117863183A, a method for locating odor sources based on adaptive spatial perception information trending, 2024-04-12, the scheme first divides the search environment into fine grids, establishes a gas diffusion model that fits the actual scene, and obtains the predicted concentration of each grid; then the robot's real-time gas sampling data is binarized, and a sensor response model based on Poisson distribution is established to obtain the sampling concentration; the robot performs Bayesian reasoning through the predicted concentration and the sampling concentration, updates the likelihood function, and thus obtains the posterior probability density function of the entire map; the reward function of each movable direction in the allowable action set is calculated through the posterior probability density function, the robot finds the direction with the largest change in the reward function, calculates the moving step length based on the current information entropy, and moves the specified step length in the direction with the largest change in the reward function. However, this method still has the following problems:

[0053] The resolution of the search space is negatively correlated with the grid size. When the resolution needs to be increased, the grid size must be reduced, and massive grids will increase the computational time and space complexity. As the search environment continues to expand, massive grids have caused the complexity of Bayesian sequential estimation and Markov decision calculation in the information trend method to increase sharply, and have become an obstacle to the application of actual robot platforms based on grid maps for information trend tracing.

[0054] To solve the above problems, the embodiments of the present invention firstly address the multi-channel information dependency problem and the robot coordination problem, and realize the intelligent allocation and execution of search and rescue tasks through the collaborative operation of drones and search robots. The drone performs real-time gas concentration monitoring and odor source tracking, while the search robot uses its collected information to plan the traceability path. The two cooperate with each other to form an efficient and accurate search and rescue system to improve the rescue speed. Secondly, in view of the problem that traditional odor tracing methods are difficult to work in real turbulent environments, the high-precision gas sensors carried by drones are combined with information trend algorithms to collect and analyze gas concentration information in the area in real time, automatically identify and track the odor diffusion source, and ensure efficient and accurate tracing. Finally, in view of the high complexity of traditional grid calculations, random particles are introduced to construct probability maps, reduce the frequency and amount of probability map updates, and achieve lightweight algorithms.

[0055] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0056] The embodiment of the present invention provides a collaborative navigation method between an odor tracing drone and a search robot, such as Figure 1 As shown, the method includes:

[0057] S1, obtain the real-time environmental information and odor data of the current location collected by the drone, and convert the environmental information into a digital map;

[0058] S2. Using the digital map and the execution parameters of the algorithm as input data, and according to the odor data of the current position, calculating the direction in which the information entropy is reduced the fastest through the information trend algorithm, and using this direction as the flight direction of the UAV during the search process; determining the odor detection point on the digital map during the flight of the UAV in the flight direction;

[0059] S3, determining whether the odor detection point is a qualified point, if so, planning a path between two qualified points through a path planning algorithm to form an initial path for the search robot, and returning to step S2;

[0060] S4. Smoothing the initial path to form a navigation path for the search robot.

[0061] The embodiment of the present invention adopts information entropy as the main basis for odor navigation, so that the search robot can still operate normally and navigate accurately in environments with poor visibility such as smoke, night or low light, thereby expanding the application scenarios of the search robot.

[0062] The following structure is Figure 2 The flowchart shown is used to describe the embodiment of the present invention in detail:

[0063] In one embodiment, S1, real-time environmental information and odor data of the current location collected by the drone are obtained, and the environmental information is converted into a digital map. The specific implementation process is as follows:

[0064] The environmental information collected by the drone through its high-precision sensors (such as cameras or odor sensors, collectively referred to as drones for the sake of convenience) is converted into a digital map.

[0065] First, the drone captures environmental data in real time and extracts key information from it. Then the acquired image or data is parsed pixel by pixel to identify the RGB value of each pixel. According to the preset threshold or classification rules, the pixels with specific RGB values ​​are classified into obstacles (such as trees, buildings, etc.), starting points (the starting position of the search and rescue mission), and end points (the location of the odor diffusion source. It should be noted that the end point is unknown at the beginning of the search, and there is no such location information in the map. The drone determines the location of the end point during the search process, and then the RGB value corresponding to the end point on the map needs to be set in advance). Finally, the classified pixel coordinates are mapped to a two-dimensional or three-dimensional space to construct an obstacle map containing obstacles, starting points, and end points. The map is presented in the form of a digital matrix or a graphical interface to facilitate the processing of the path planning algorithm.

[0066] like Figure 3 As shown, the dark gray part in the visualized digital map is the obstacle, the yellow point is the starting point, and the green point is the end point. The constructed digital map will serve as a prerequisite for the robot to plan and analyze the path.

[0067] In one embodiment, S2, using the digital map and the execution parameters of the algorithm as input data, and calculating the direction in which the information entropy is reduced the fastest through the information trend algorithm according to the odor data at the current location, as the flight direction of the UAV during the search process; determining the odor detection points on the digital map during the flight direction of the UAV.

[0068] The digital map information and algorithm-related parameters are input, and the direction in which the information entropy decreases the fastest is calculated based on the information entropy distribution of the current position and surrounding environment detected by the drone. In each step of the search, the drone moves a certain distance in that direction and updates its position information in real time. The information tendency algorithm is a source search method that obtains the local maximum information gain based on information tendency, that is, the odor source is estimated using the sequentially updated probability distribution, and the convergence degree of the probability distribution is measured by Shannon entropy. The drone moves toward the source in the convergence direction of the probability distribution (that is, the maximum entropy drop, also known as information gain). At the same time, the gas concentration data collected by the drone is used to mark the odor detection points along the way during the search process. For the obtained odor points, it is necessary to detect whether they overlap with the obstacle map. If there is an overlap, the detection point is discarded, and if there is no overlap, it is retained.

[0069] The odor point detected by the drone in the air may be blocked by obstacles on the ground, so the search robot cannot completely follow the route of the drone in the air to the odor source, so the route of the drone needs to be optimized to obtain the route of the search robot. The information trend algorithm is used by drones to detect and track the odor source. The drone is used to track the odor source, and the search robot optimizes the best route based on the drone's route and actual terrain.

[0070] The specific process is as follows:

[0071] S201, using the digital map and the execution parameters of the algorithm as input data, and calculating the information entropy distribution through the information trend algorithm according to the odor data of the current location, specifically including:

[0072] The basis of the information trend algorithm is that the searcher has the ability to predict the concentration of particles released by surrounding emission sources, that is, to predict the concentration of particles released by emission sources at different locations in space through the gas diffusion model. The general gas diffusion model satisfies the following formula:

[0073]

[0074] Wherein, V represents the average wind speed in m / s; is the Hamiltonian operator; C(r|r s ) is the average concentration of odor particles at position r; D refers to the effective isotropic diffusion coefficient in the turbulence model, with the unit of m 2 / s; τ refers to the average life span of particles released from the release source, in seconds; R refers to the release rate of the release source; δ is the Kronecker function;

[0075] The information state uses the probability distribution P(r s |L l:k ) indicates that r s Refers to the location of the leak source. l:k = {Z i (r i )} l≤i≤k Refers to the collection of all measurements of the drone from the beginning to time. The posterior probability distribution is updated using the Bayesian framework, and the update process is shown in the following formula:

[0076]

[0077] Information entropy can be used to describe the degree of uncertainty in the location of the release source. When uncertainty increases, entropy also increases, which is the core concept of the information orientation algorithm. The algorithm relies on the Bayesian equation to update the release source probability map and calculates the change in information entropy by probability to determine the movement direction of the mobile searcher. The expression of information entropy at time t is shown as follows:

[0078] S t =-∑P(r s |L l:k )logP t (r s |L l:k )

[0079] S202, determining the direction in which the information entropy decreases the fastest as the flight direction of the UAV during the search process, specifically including:

[0080] Predict the information entropy of the next alternative moving position, and select the direction where the information entropy decreases fastest as the starting position of the next moving position. The information entropy subtraction equation is as follows:

[0081] ΔS full =

[0082] P s_found (r s |L l:k )(-S t )+

[0083] P s_not_found (r s |L l:k )ΔS s_not_fo

[0084] Among them, P s_found Refers to the probability P of finding the release source on the probability graph s_not_found Refers to the probability and probability of not finding the release source on the probability graph; S t Refers to the decrease in information entropy when the release source is found. When the release source is found in an ideal situation, the information entropy is 0. Refers to the reduction of information entropy without finding the source of release.

[0085] S203, introduce Monte Carlo random sampling to determine the odor detection points on the digital map during the flight of the drone in the flight direction. Specifically including:

[0086] Traditional information trending algorithms usually construct grid maps to represent probability distributions. However, the massive grids make the complexity of Bayesian sequential estimation and Markov decision calculations in information trending methods increase dramatically, and the movement of the search robot is limited to adjacent grids, which makes it difficult to show the robot's motion performance and reduces the traceability efficiency. Therefore, the embodiment of the present invention introduces Monte Carlo random sampling for the representation of probability maps, that is, using random sampling points in the state space to approximate the posterior probability distribution of random variables. If M sampling points are randomly selected According to the Monte Carlo method, the posterior probability distribution P(X 0:k |Y 1:k ) is approximately:

[0087]

[0088] Among them, X k is the sampling point distribution at time k, is the position of the mth particle at time k, Y kis the state observation value at time k, and δ(·) represents the Dirac function. Thus, the probability map of the odor source distribution can be constructed.

[0089] After the probability map is constructed, the search robot is set to follow the search route of the drone and start moving in the map from the starting point. When the search robot moves to a certain position r, the position r of the plume source in the gas diffusion model is combined 0 Calculate the expected hit rate R(r|r 0 ), then the expected average number of hits per unit time Δt is:

[0090] k=ΔtR(r|r 0 )

[0091] This constructs a Poisson distribution model:

[0092]

[0093] The hit is simulated by judging whether the Poisson distribution random number z is greater than 0. If the smell is detected, the detection point is recorded, otherwise, the information entropy distribution is recalculated until the search stops at the location of the smell source.

[0094] In one embodiment, S3, determine whether the odor detection point is a qualified point, if so, plan a path between two qualified points through a path planning algorithm to form an initial path for the search robot, and return to step S2. The specific implementation process is as follows:

[0095] It should be noted that the embodiment of the present invention adopts the RRT (Rapidly-Exploring Random Tree) algorithm through the path planning algorithm. In the specific implementation process, other path planning algorithms can also be used to plan the path between two odor detection points, such as the probabilistic road map algorithm (PRM), the A* algorithm, the ant colony algorithm, etc.

[0096] The RRT (Rapidly-Exploring Random Tree) algorithm is a sampling-based global path planning algorithm that can be used in multi-dimensional space. In the embodiment of the present invention, when the search robot obtains each odor detection point, it will determine its distance from the previous odor detection point and the distance from the obstacle. If it is too close to the previous odor detection point or overlaps with the position of the obstacle, the point does not meet the requirements. If the odor detection point obtained meets the requirements, it is called a qualified point (the starting point of the search robot and the odor diffusion source are also qualified points). When the drone obtains each qualified point, it will pass its coordinates and the coordinates of the previous qualified point into the corresponding obstacle map and plan the path between the two points through the RRT algorithm.

[0097] At the same time, the path distance threshold γ is set, which is equal to the odor detection point a detected by the drone for the third time. 3 The path distance h from the previous monitoring point 3 After setting the threshold, the nth detection point a obtained by the drone n If the path distance between the previous detection point and n If it is less than the threshold γ, the detection point a is ignored. n , then return to step S2 and reacquire new odor detection points until the distance between the two odor detection points is greater than the threshold γ, and then use them as qualified points for the next step of path optimization.

[0098] In one embodiment, S4, the initial path is smoothed to form a navigation path for the search robot. The specific implementation process is as follows:

[0099] When the path between two qualified points is determined, the obtained path is input into the optimization algorithm for processing. To further improve the smoothness and safety of the path, the algorithm uses the path smoothing technology based on B-spline curves to perform refined smoothing on the preliminary route. It should be noted that in the specific process, other algorithms can also be used to smooth the path, such as the Douglas-Peucker algorithm.

[0100] The embodiment of the present invention uses the De Boor algorithm to control node generation, and the node vectors are arbitrarily distributed to form a non-uniform B-spline curve. The De Boor algorithm is an efficient algorithm used in the B-spline curve theory to calculate the value of the B-spline basis function and its derivatives. Traditionally, the node vectors of the B-spline curve are uniformly distributed, which limits the diversity of the curve shape. The embodiment of the present invention allows the node vectors to be freely distributed in the parameter space by introducing the De Boor algorithm, that is, the nodes can be arranged non-uniformly, which makes it possible to accurately adjust the curve shape. The De Boor algorithm recursively calculates the value of each basis function at any given parameter value, so that even if the node distribution is complex, the points on the curve can be calculated quickly and accurately. The basis function of the B-spline curve is defined by the Cox-de Boor recursion:

[0101]

[0102] Among them, t is the node set, N i,k (t) is called the k-th (k+1-th order) B-spline basis function of the i-th control node, i=0,1,,n-1,k≥1 and 0 / 0=0.

[0103] By controlling the distribution of node vectors and the order of the B-spline curve, sharp turns and unnecessary broken lines in the path can be effectively reduced, thereby generating a smoother and shorter path as the navigation path of the search robot.

[0104] In the specific implementation process, it also includes S5, obtaining the data after the search robot searches along the navigation path, and judging whether the search robot reaches the odor source. If so, the search is terminated, otherwise, it returns to step S2.

[0105] The search robot searches along the navigation path to determine whether it has reached the odor source. If so, the search ends. Otherwise, the search returns to step S2 and steps S2 to S5 are re-executed to reach the odor source. The final search path is as follows: Figure 4 shown.

[0106] It should be noted that the above three steps can be executed by the same computer, and the information in the drone and the search robot can be transmitted to the computer through the wireless network. The computer will perform unified calculations to form the navigation path of the search robot, and then transmit the navigation path to the search robot through the wireless network. The drone and the search robot can also have their own processors, and information is transmitted between the two through the wireless network, and the navigation path is formed by the built-in processors. In the above embodiment, the process of generating the navigation path is described by executing the same computer. The transmission of data belongs to the prior art and is not described in detail in the above embodiment. If the navigation path is formed by the processors built into the drone and the search robot, the execution subject can be replaced with their respective processors, which will not be repeated here.

[0107] An embodiment of the present invention provides a collaborative navigation system of an odor tracing drone and a search robot, comprising:

[0108] The data acquisition module is used to execute S1, obtain the real-time environmental information collected by the drone and the odor data of the current location, and convert the environmental information into a digital map;

[0109] The odor detection point determination module is used to execute S2, take the digital map and the execution parameters of the algorithm as input data, calculate the direction in which the information entropy is reduced the fastest according to the odor data of the current position through the information trend algorithm, and use it as the flight direction of the UAV during the search process; determine the odor detection point on the digital map during the flight of the UAV according to the flight direction;

[0110] A path planning module is used to execute S3, determine whether the odor detection point is a qualified point, and if so, plan a path between two qualified points through a path planning algorithm to form an initial path of the search robot, and return to step S2;

[0111] The path smoothing module is used to execute S4, smooth the initial path, and form a navigation path for the search robot.

[0112] It is understandable that the odor-tracing drone and search robot collaborative navigation system provided in the embodiment of the present invention corresponds to the above-mentioned odor-tracing drone and search robot collaborative navigation method, and the explanations, examples, beneficial effects and other parts of its relevant contents can refer to the corresponding contents in the odor-tracing drone and search robot collaborative navigation method, which will not be repeated here.

[0113] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program for collaborative navigation of an odor-tracing drone and a search robot, wherein the computer program enables a computer to execute the collaborative navigation method of an odor-tracing drone and a search robot as described above.

[0114] An embodiment of the present invention also provides an electronic device, comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the programs include a method for executing the collaborative navigation method of the odor tracing drone and the search robot as described above.

[0115] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0116] 1. Different from the traditional navigation method that relies on visual information, the embodiment of the present invention uses information entropy as the main basis for odor navigation, so that the search robot can still operate normally and navigate accurately in environments with poor visibility such as smoke, night or low light, expanding the application scenarios of the search robot

[0117] 2. The embodiment of the present invention realizes the real-time collaboration between the drone information collection and the search robot. The drone transmits the collected map and odor information to the search robot in real time. The search robot processes the obtained data to plan and analyze the optimal path to the next path point, ensuring that the search and rescue operation in the real complex environment can avoid being affected by environmental changes and carry out the search and rescue operation in a timely and accurate manner.

[0118] 3. In the embodiment of the present invention, random sampling particles are introduced to replace the grid map in the information trending algorithm, thereby reducing the computational complexity when making decisions on movement, achieving lightweight algorithms, and removing the constraints on the robot's movement direction, thereby fully utilizing the mobility of the search robot and improving search efficiency.

[0119] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0120] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A collaborative navigation method for an odor tracing drone and a search robot, characterized in that: include: S1, obtain the real-time environmental information and odor data of the current location collected by the drone, and convert the environmental information into a digital map; S2. Using the digital map and the execution parameters of the algorithm as input data, and according to the odor data of the current position, calculating the direction in which the information entropy is reduced the fastest through the information trend algorithm, and using this direction as the flight direction of the UAV during the search process; determining the odor detection point on the digital map during the flight of the UAV in the flight direction; S3, determining whether the odor detection point is a qualified point, if so, planning a path between two qualified points through a path planning algorithm to form an initial path for the search robot, and returning to step S2; S4, smoothing the initial path to form a navigation path for the search robot; Wherein, the S2 specifically includes: S201, using the digital map and the execution parameters of the algorithm as input data, and calculating the information entropy distribution through the information trend algorithm according to the odor data of the current location; S202, determining the direction in which the information entropy decreases fastest as the flight direction of the UAV during the search process; S203, introducing Monte Carlo random sampling to determine the odor detection points on the digital map during the flight of the drone in the flight direction, including: Introducing Monte Carlo random sampling, if M sampling points are randomly selected According to the Monte Carlo method, the posterior probability distribution P(X 0:k |Y 1:k ) is approximately: Among them, X k is the sampling point distribution at time k, is the position of the mth particle at time k, Y k is the state observation value at time k, and δ(·) represents the Dirac function; thus, the probability map of the odor source distribution can be constructed; After constructing the probability map, the search robot is set to follow the search route of the drone and start moving in the map from the starting point. When the search robot moves to a certain position r, the expected hit rate R(r|r0) is calculated in combination with the position r0 of the plume source in the gas diffusion model. The expected average hit number per unit time Δt is: k=ΔtR(r|r0) This constructs a Poisson distribution model: The hit is simulated by judging whether the Poisson distribution random number z is greater than 0, that is, whether the smell is detected; if the smell is detected, the detection point is recorded, otherwise, the information entropy distribution is recalculated until the search stops at the location of the smell source.

2. The method for collaborative navigation of an odor tracing drone and a search robot according to claim 1, characterized in that: The step of judging whether the odor detection point is a qualified point includes: Determine whether the odor detection point overlaps with the position of the obstacle, or whether the distance between the current odor detection point and the previous odor detection point is less than or equal to the threshold; if one of the above conditions is met, it is an unqualified point; if neither condition is met, it is a qualified point.

3. The collaborative navigation method of the odor tracing drone and the search robot according to claim 1, characterized in that: The path planning algorithm includes a fast traversal random tree algorithm.

4. The method for collaborative navigation of an odor tracing drone and a search robot according to claim 1, characterized in that: The smoothing of the initial path includes: using a path smoothing technology based on a B-spline curve to smooth the initial path.

5. The collaborative navigation method of an odor tracing drone and a search robot according to any one of claims 1 to 4, characterized in that: Also includes: S5. Obtain the data after the search robot searches along the navigation path, and determine whether the search robot has reached the odor source. If so, end the search; otherwise, return to step S2.

6. A collaborative navigation system of an odor tracing drone and a search robot, characterized in that: include: The data acquisition module is used to execute S1, obtain the real-time environmental information collected by the drone and the odor data of the current location, and convert the environmental information into a digital map; The odor detection point determination module is used to execute S2, take the digital map and the execution parameters of the algorithm as input data, calculate the direction in which the information entropy is reduced the fastest according to the odor data of the current position through the information trend algorithm, and use it as the flight direction of the UAV during the search process; determine the odor detection point on the digital map during the flight of the UAV according to the flight direction; A path planning module is used to execute S3, determine whether the odor detection point is a qualified point, and if so, plan a path between two qualified points through a path planning algorithm to form an initial path of the search robot, and return to step S2; A path smoothing module is used to execute S4 and smooth the initial path to form a navigation path for the search robot; Wherein, the S2 specifically includes: S201, using the digital map and the execution parameters of the algorithm as input data, and calculating the information entropy distribution through the information trend algorithm according to the odor data of the current location; S202, determining the direction in which the information entropy decreases fastest as the flight direction of the UAV during the search process; S203, introducing Monte Carlo random sampling to determine the odor detection points on the digital map during the flight of the drone in the flight direction, including: Introducing Monte Carlo random sampling, if M sampling points are randomly selected According to the Monte Carlo method, the posterior probability distribution P(X 0:k |Y 1:k ) is approximately: Among them, X k is the sampling point distribution at time k, is the position of the mth particle at time k, Y k is the state observation value at time k, and δ(·) represents the Dirac function; thus, the probability map of the odor source distribution can be constructed; After constructing the probability map, the search robot is set to follow the search route of the drone and start moving in the map from the starting point. When the search robot moves to a certain position r, the expected hit rate R(r|r0) is calculated in combination with the position r0 of the plume source in the gas diffusion model. The expected average hit number per unit time Δt is: k=ΔtR(r|r0) This constructs a Poisson distribution model: The hit is simulated by judging whether the Poisson distribution random number z is greater than 0, that is, whether the smell is detected; if the smell is detected, the detection point is recorded, otherwise, the information entropy distribution is recalculated until the search stops at the location of the smell source.

7. A computer-readable storage medium, characterized in that: It stores a computer program for collaborative navigation between an odor tracing drone and a search robot, wherein the computer program enables a computer to execute the collaborative navigation method between an odor tracing drone and a search robot as described in any one of claims 1 to 5.

8. An electronic device, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the programs include a method for executing the collaborative navigation method of the odor tracing drone and the search robot as described in any one of claims 1 to 5.

Images