Real-time path planning method fusing improved locust-dynamic window approach

Abstract

The application discloses a real-time path planning method fusing improved locust-dynamic window method, and comprises the following steps: S1, generating an initial path according to an improved A* algorithm, performing global path planning on the path in a radiation environment through an improved locust optimization algorithm, and determining the path from the starting point to the ending point; S2, performing local path planning according to the path through an improved DWA algorithm, and generating a planning path. Through the introduction of a position updating strategy considering radiation distribution information and an improved adaptive comfort zone coefficient, the influence of the radiation risk on path selection is more truly reflected, and the position updating process of the GOA is optimized. This enables the algorithm to effectively guide the locust population to avoid high radiation areas in a nuclear environment, and significantly reduces the radiation exposure risk. Through the combination of the global optimization capability of the improved A* algorithm, the initial population distribution of the locust optimization algorithm is improved, and then the convergence speed of the locust optimization algorithm is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of path planning, and particularly relates to a real-time path planning method fusing an improved locust-dynamic window method. BACKGROUND

[0002] The position of nuclear energy in the global energy structure continues to rise, and the scale of nuclear facility construction is expanding. However, the nuclear safety problem that follows is also increasingly serious, especially when a sudden accident occurs, higher requirements are put forward for emergency response capability. Under this background, robots are widely used in nuclear emergency tasks to reduce the risk of radiation exposure of personnel and undertake operation tasks that are difficult for humans to complete. However, when the robot performs a task in a radioactive environment, it often faces the risk of high radiation accumulation, which easily leads to damage to its electronic components and affects the execution of the task. At the same time, after a nuclear accident, the environmental state usually has high uncertainty, which further increases the complexity of task execution. Therefore, in order to ensure that the robot can successfully complete the emergency task, it is particularly important to respond to environmental changes in real time and plan a path that effectively reduces the accumulated radiation dose of the robot.

[0003] Path planning technology is a core link for ensuring that a robot can efficiently and reliably perform a task in a complex environment. In a radiation environment, common path planning methods mainly include traditional graph search methods, sampling-based methods, and heuristic intelligent optimization methods. The traditional graph search method represented by the A* algorithm is prone to a sharp increase in computational complexity in high-dimensional space or large-scale environment, and its processing time and memory consumption increase exponentially. At the same time, this kind of method needs to discretize the continuous space, which easily causes the loss of environmental details and leads to the generation of suboptimal paths. The sampling method represented by the Rapidly-exploring Random Tree (RRT) has certain advantages in search efficiency and is suitable for high-dimensional and complex spaces, but its inherent randomness leads to insufficient result stability, slow convergence in narrow channels or highly constrained environments, and easy generation of non-optimal paths, which often need to be optimized by post-processing to improve performance. The heuristic intelligent optimization method represented by genetic algorithm and particle swarm optimization has strong global search ability and can jump out of the local optimum to a certain extent, but its convergence performance is highly sensitive to parameter settings, lacks global optimality guarantee, and has large computational overhead, which is difficult to meet the real-time requirement.

[0004] Locust optimization algorithm is an intelligent optimization method inspired by the migration and foraging rules of locust swarms in nature, aiming to solve complex optimization problems. Compared with traditional swarm intelligence algorithms such as genetic algorithm and particle swarm optimization, this method emphasizes more on the cooperation and interaction between populations, has strong global search ability, and helps to avoid local optimal trap, especially in handling multi-peak function and composite optimization tasks. The algorithm can efficiently explore and find solutions close to the global optimum in the given search space, and through dynamic adjustment of parameters, it realizes the reasonable balance between exploration and development stages, thereby improving the overall search performance. On standard test functions, its results usually show lower mean and variance, and in practical applications such as engineering design, it shows high solving efficiency. In addition, locust optimization algorithm has a simple structure, is easy to extend and improve, and does not need to rely on gradient information. Its unique position updating strategy considers the positions of all individuals in the group, continuously corrects the search behavior and movement direction through information interaction between individuals, thereby enhancing the search space coverage and optimization accuracy, making it have faster convergence characteristics and higher reliability in complex environments.

[0005] Although existing algorithms have improved the efficiency of path planning and the control of cumulative radiation dose to some extent, they have not fully considered the impact of uncertainty in the radiation environment on path planning. In the event of a nuclear accident or emergency, radiation information is often incomplete or uncertain, which poses unpredictable challenges to the path planning process. In addition, existing methods usually only focus on the single factor of radiation dose to calculate the actual cost of the path, ignoring multiple constraint conditions in path planning. In a complex nuclear radiation environment, the length of the path may directly affect the energy consumption of the robot, and the energy consumption may affect the work capacity of the robot within a certain time, thereby affecting the efficiency of task completion. Therefore, in practical applications, path planning not only needs to consider the minimization of radiation dose, but also needs to consider factors such as path length. SUMMARY

[0006] In view of the above problems in the prior art, the present application provides a real-time path planning method based on improved locust-dynamic window method, which effectively realizes efficient and safe real-time path planning of robots in complex radiation environments, and solves the problems that existing path planning methods cannot obtain the minimum cumulative dose path and respond to unknown environments in a radiation environment with uncertainty.

[0007] To achieve the above-mentioned purposes, the technical scheme adopted by the present application is as follows: a real-time path planning method based on improved locust-dynamic window method, comprising the following steps:

[0008] S1, generating an initial path according to the improved A* algorithm, performing global path planning on the path in the radiation environment by introducing an improved locust optimization algorithm, and determining the path from the starting point to the ending point;

[0009] S2, generating a planning path according to the path by the improved DWA algorithm.

[0010] Further, S1 comprises the following steps:

[0011] S11, generating an initial path by using the improved A* algorithm to obtain the initial position of the population; calculating the fitness value of each locust individual in the population, selecting the optimal fitness value as the target fitness value; initializing the number of locusts and the current iteration number, and setting the maximum iteration number;

[0012] S12, setting the number of locusts to 1;

[0013] S13, updating the population position according to the position update strategy considering the radiation distribution information;

[0014] S14, calculating the current fitness value according to the current population position, and judging whether the current fitness value is better than the target fitness value; if yes, updating the target fitness value and the target optimal solution position according to the current fitness value and the current population position, and entering S15; if no, entering S15;

[0015] S15, judging whether the set population number is reached, if yes, entering S16, if no, setting the current number of locusts to one more, and returning to S13;

[0016] S16, updating the population position according to the hybrid teaching-learning optimization algorithm;

[0017] S17, judging whether the current iteration number reaches the maximum iteration number, if yes, outputting the target optimal solution position to generate the path from the starting point to the ending point; if no, setting the current iteration number to one more, and returning to S12.

[0018] Further, in S11, the improved A* algorithm optimizes the estimation process of (f) by introducing a multi-scale evaluation method, and the optimized expression of (f) is specifically: H n H n

[0019]

[0020] In the formula, w is a weight factor, λ i is the distance between adjacent two points, L is the local radiation dose of the +1th node, a is the radiation dose rate of the th node, a is the radiation dose of the th node, a ​​​​The radiation dose rate of the +1th node, v t For the moving speed, N For the number of points between the current point and the end point;

[0021]

[0022] In the formula, The weight of the first a The radiation dose of the first b Adjacent point around the +1th node, The weight of the first b Point.

[0023] Further: in S13, the position update strategy considering the radiation distribution information calculates the expression of the position of the locust i The expression is as follows:

[0024]

[0025] In the formula, b max The upper bound of the search space, b min The lower bound of the search space, T d The best value in the target, d ij The distance between the locust i And j The unit vector from the position of the locust To the position of the locust i , j The gradient weight, The radiation gradient of the current position in dimension , estimated by finite difference method, d The population number, N The adaptive comfort zone coefficient, c The current iteration number, n The maximum iteration number, L The function for calculating the interaction force between locust populations, whose expression is as follows:

[0026]

[0027] In the formula, f The attraction strength parameter, l The attraction scale parameter.

[0028] Further: in S13, the adaptive comfort zone coefficient c Adaptively adjusts according to the radiation intensity, and its expression is as follows:​​

[0029]

[0030] In the formula, c max This represents the maximum value of the adaptive comfort zone coefficient. c min This represents the minimum adaptive comfort zone coefficient. α As a regulating factor; For the first i Only the locust's current location is radiating. R min This represents the minimum radiation value on the map. R max This represents the maximum radiation value on the map.

[0031] Furthermore: S16 includes the following sub-steps:

[0032] S161. Update the population position during the teacher stage. The specific position update formula is as follows:

[0033]

[0034] In the formula, locusts i new location, locusts i The original location, rand The random number is a uniformly distributed number within the range (0,1). X Teacher For the optimal individual, β As a teaching factor, Mean It is the population mean;

[0035] S162. Update the population position during the learner phase. The specific formula for position update is as follows:

[0036]

[0037] In the formula, Locusts selected randomly k The original location, The fitness of individual locusts.

[0038] Furthermore, S2 includes the following sub-steps:

[0039] S21. Extract the optimal path node from the path node sequence of the path, and take the starting point as the current point;

[0040] S22. Set the next key point of the current point as the local sub-target point;

[0041] S23. Perform velocity sampling and select the optimal path based on the improved DWA trajectory evaluation function;

[0042] S24, determining whether the local sub-target point is reached, if not, returning to S23, if yes, entering S25;

[0043] S25, determining whether the local sub-target point is the end point, if not, returning to S22, if yes, outputting the planning path from the start point to the end point.

[0044] Further, in S23, the improved DWA trajectory evaluation function The expression is specifically:

[0045]

[0046] In the formula, is the radiation evaluation function, is the azimuth angle evaluation function, is the distance between the robot at the end point of the predicted trajectory and the nearest obstacle on the map, is the linear velocity and angular velocity of the current robot, v , ω is the velocity space, v is the linear velocity, ω is the angular velocity, , , , and is the weight coefficient of the DWA algorithm;

[0047]

[0048] In the formula, is the average radiation value of each sampling point on the trajectory;

[0049]

[0050] In the formula, is the radiation intensity value of the m sampling point on the trajectory, M is the number of sampling points on the trajectory.

[0051] Further, the weight coefficient heading ( v , ω ) is adjusted according to the distance between the robot and the obstacle The expression of the adjusted weight coefficient is specifically:

[0052]

[0053] In the formula, P is the adaptive weight distance threshold value of the robot;

[0054] Adjusting the weight coefficient of the linear velocity and the angular velocity of the robot velocity v ω The expression of the adjusted weight coefficient

[0055]

[0056] In the formula, ω smooth is the smoothness weight;

[0057]

[0058] In the formula, v Delta is the change amount of the linear velocity, ω Delta is the change amount of the angular velocity.

[0059] The beneficial effects of the present application are:

[0060] (1) The present application more truly reflects the influence of radiation risk on path selection by introducing a position update strategy considering radiation distribution information and an improved adaptive comfort zone coefficient, and optimizes the position update process of GOA. This enables the algorithm to effectively guide the locust population to avoid high radiation areas in a nuclear environment, significantly reducing the risk of radiation exposure.

[0061] (2) The present application improves the initial population distribution of the locust optimization algorithm by combining the global optimization ability of the improved A* algorithm, thereby improving the convergence speed of the locust optimization algorithm. This reduces the time required for path search and effectively reduces the radiation dose on the path.

[0062] (3) The present application improves the initial population distribution of the locust optimization algorithm by combining the global optimization ability of the improved A* algorithm, thereby improving the convergence speed of the locust optimization algorithm. This reduces the time required for path search and effectively reduces the radiation dose on the path.

[0063] (4) By combining the improved locust optimization algorithm and the improved DWA algorithm, real-time path planning in complex radiation environments can be effectively realized. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 The present application is a fusion improved locust-dynamic window method real-time path planning method flow chart.

[0065] Figure 2 The present application is a path planning flow chart.

[0066] Figure 3 ​​​​For multi-scale evaluation in the A* algorithm of the present application Figure 1 .

[0067] Figure 4 For multi-scale evaluation in the A* algorithm of the present application Figure 2 . DETAILED DESCRIPTION

[0068] The specific embodiments of the present application are described below to facilitate the understanding of the present application for those skilled in the art, but it should be clear that the present application is not limited to the scope of the specific embodiments, and for those skilled in the art, it is obvious that all the inventions utilizing the concept of the present application are within the spirit and scope of the present application as defined and determined by the appended claims.

[0069] As Figure 1 shown, in one embodiment of the present application, the real-time path planning method combining the improved locust-dynamic window method includes the following steps:

[0070] S1, generating an initial path according to the improved A* algorithm, performing global path planning on the path in the radiation environment by introducing the improved locust optimization algorithm, and determining the path from the starting point to the ending point;

[0071] S2, performing local path planning according to the path by the improved DWA algorithm to generate a planning path.

[0072] The basic idea of the present application is to introduce multi-level improvement in the locust optimization algorithm to improve its path planning performance in the radiation environment. First, the improved A* algorithm is used to initialize the locust population, and then the teaching-learning optimization algorithm is combined in the position updating mechanism, and the radiation gradient term and the adaptive comfort zone coefficient c are introduced to speed up the convergence speed and enhance the adaptability to the dynamic changes of the radiation field. Secondly, after obtaining the global path based on the improved locust optimization algorithm, the improved DWA algorithm is combined for local path planning, so that the generated path not only has the characteristic of minimum cumulative radiation dose, but also can respond to the dynamic changes of unknown obstacles and radiation dose in real time.

[0073] S1 includes the following steps:

[0074] S11, generating an initial path by using the improved A* algorithm to obtain the initial position of the population; calculating the fitness value of each locust individual in the population, selecting the optimal fitness value as the target fitness value; initializing the number of locusts and the current iteration number, and setting the maximum iteration number;

[0075] S12, let the number of locusts be 1;

[0076] S13, updating the population position according to the position updating strategy considering the radiation distribution information;

[0077] S14, calculating a current fitness value according to the current population position, judging whether the current fitness value is better than a target fitness value; if yes, updating the target fitness value and a target optimal solution position according to the current fitness value and the current population position, and entering S15; if no, entering S15;

[0078] S15, judging whether a set population quantity is reached; if yes, entering S16; if no, letting a current locust quantity be one more, and returning to S13;

[0079] S16, updating the population position according to the hybrid teaching-learning optimization algorithm;

[0080] S17, judging whether a current iteration number reaches a maximum iteration number; if yes, outputting the target optimal solution position, and generating a path from a starting point to a terminal point; if no, letting the current iteration number be one more, and returning to S12.

[0081] In the embodiment, a path planning flow chart is as shown in Figure 2 The path planning problem of the robot in the radiation environment is solved by introducing the multi-level improved locust optimization algorithm, and finally a feasible and optimized complete path from the starting point to the target point is determined.

[0082] The locust optimization algorithm is an optimization algorithm designed according to the migration and foraging behavior of locusts in nature. The algorithm uses the interaction between individuals to search for the best food source. As a heuristic algorithm, the performance of the algorithm is sensitive to the initial distribution of the population. The traditional locust optimization algorithm usually uses a random initialization method to determine the population position, but this method may be difficult to approach the optimal path in time in the early search, thereby delaying the convergence speed. In order to shorten the early search time and improve the search efficiency, the method of the present application introduces the path generated by the improved A* algorithm in the population initialization stage, so that part of the individuals are distributed in the vicinity of the global optimal solution, thereby focusing on the high-quality solution space in the early search, accelerating the convergence and improving the overall optimization performance. In the A* algorithm, the selection of H(n) plays a crucial role in the search path of the A* algorithm. Therefore, the improved A* algorithm optimizes the estimation process of H(n) by introducing a multi-scale evaluation method. Through local scale estimation, the radiation situation near each point on the path from the current node to the terminal point is evaluated. At the same time, the radiation cost of the entire path is evaluated through global scale evaluation, and the cumulative radiation dose from the current node to the terminal point is calculated, as shown in Figure 3 and Figure 4 Finally, the global and local costs are weighted and fused to obtain the final estimated cost.

[0083] In S11, the improved A* algorithm optimizes the estimation process of H ( n ) by introducing a multi-scale evaluation method, and the optimizedH ( n The specific expression for ) is:

[0084]

[0085] In the formula, λ As a weighting factor, L i The distance between two adjacent points. For the first a The local radiation dose of node +1 is calculated by weighted averaging to obtain the dose of node 1. a +1 local cost of the 7 neighboring points around the node, For the first a Radiation dose rate at each node For the first a Radiation dose rate of +1 node v t For movement speed, N This represents the number of points between the current point and the destination.

[0086]

[0087] In the formula, For the first a +1 node surrounding the first b Radiation dose at adjacent points For the first b The weight of each point.

[0088] In calculating the fitness value of each locust individual in the population, this invention designs a comprehensive evaluation function that considers both path length to ensure search efficiency and cumulative radiation dose constraints to reflect safety under nuclear radiation conditions. This weighted combination of the two aspects makes the fitness assessment more aligned with the actual needs of nuclear emergency missions, rather than relying solely on the traditional single path length indicator. The optimized objective function... f The specific expression is:

[0089]

[0090] In the formula, ω 1 and ω 2 represents the path length cost weight and the radiation dose cost weight. K The number of points on the path. and These are the coordinates of two adjacent points on the path. For the first p Radiation dose rate at each node For the first p Radiation dose rate of +1 node v tFor the moving speed, L i For the distance between two adjacent points.

[0091] In S13, the position updating strategy considering the radiation distribution information is calculated i The position The expression is specifically:

[0092]

[0093] In the formula, b max The upper bound of the search space, b min The lower bound of the search space, T d The best value in the target, d ij The distance between the locust i And j The unit vector from the position of the locust To the position of the locust i , j The gradient weight, The radiation gradient of the current position in dimension Is estimated by finite difference method, d The population number, N The adaptive comfort zone coefficient, c The current iteration number, n The maximum iteration number, L The function for calculating the interaction force between locust populations, the expression is specifically:

[0094]

[0095] In the formula, f The attraction strength parameter, l The attraction scale parameter.

[0096] In this embodiment, the position updating formula of the locust optimization algorithm is improved, and the principle of the position updating strategy considering the radiation distribution information is established, which is specifically:

[0097] In the locust optimization algorithm, each locust individual represents a candidate solution of the optimization problem, and the locust individual is mainly affected by three factors: the interaction force between locust populations, wind force and gravity. The mathematical expression for simulating the behavior of locust swarm is as follows:

[0098]

[0099] In the formula, The locust​i Location, S i locusts i Influenced by inter-population interactions, G i locusts i The influence of external gravity, A i locusts i The influence of external wind force. In solving mathematical optimization problems, in order to optimize the mathematical model, External gravitational influence term in the formula G i Wind force impact item A i It needs to be replaced with the optimal value in the target. T d We get the following formula:

[0100]

[0101] in: S i Calculated using the following formula:

[0102]

[0103] However, this mathematical model cannot be directly used for optimization. This is because locusts quickly enter their comfort zone during the search process, making it difficult for them to converge to the specified target point; furthermore, the model does not incorporate radiation distribution information, resulting in a lack of proactive avoidance ability in high-radiation areas. Therefore, for... The formula is further rewritten to both guide individuals to converge to the target position and avoid high-radiation areas during the search process, thus better meeting the path optimization requirements in a radiation environment, thereby obtaining... The expression.

[0104] In the standard locust optimization algorithm, the adaptive comfort zone coefficient c typically employs a linear decay strategy, meaning that as the number of iterations n increases, c decreases from its maximum value cn. max Gradually decrease to the minimum value c min This approach aims to balance the algorithm's early global exploration capabilities with its later local planning capabilities. While this strategy enables a shift from global to local search, it doesn't consider the non-uniform distribution of the radiation field. Therefore, in high-radiation areas, individuals may become trapped in local optima or fail to effectively avoid radiation hotspots. To enhance the algorithm's adaptability to the radiation environment, this invention proposes an adaptive c-coefficient adjustment mechanism based on radiation intensity. The core idea is to couple the value of c with the radiation intensity at the locust's current location, thereby dynamically adjusting the individual's update step size to proactively avoid high-radiation areas. (Adaptive comfort zone coefficient) cAccording to the adaptive adjustment of the radiation intensity, the expression is specifically as follows:

[0105]

[0106] In the formula, c max is the maximum value of the adaptive comfort zone coefficient, c min is the minimum value of the adaptive comfort zone coefficient, and in the embodiment c max and c min are respectively 1 and 0.0004, α is the adjustment factor; is the first i is only the radiation of the current position of the locust, R min is the minimum value of the map radiation, R max is the maximum value of the map radiation. Based on the strategy, when the locust is located in a high radiation zone, the adjustment term increases c the value, promotes the exploration behavior of a larger step, and helps the population to quickly escape from a high-risk area; on the contrary, in a low radiation zone, the adjustment term reduces c the value, and strengthens the local search to optimize the path length.

[0107] In the whole search iteration process, the evaluation index of the pros and cons of each position of the locust is the fitness function (objective function). For the path planning of the mobile robot in the nuclear radiation environment, it is usually required to minimize the cumulative radiation dose as the fitness function. In the solving process, the population position is constantly updated by considering the position update strategy of the radiation distribution information to obtain the optimal solution of the problem. The best fitness obtained after each iteration is recorded as the optimal solution of the current problem, until the maximum number of iterations is reached.

[0108] In S16, the improved locust algorithm realizes global optimization by simulating the social interaction of locusts, but may be trapped in local optimum or slow convergence in a complex radiation environment. In order to improve the search efficiency and diversity of the algorithm, the locust algorithm is combined with the teaching-learner optimization algorithm to form a hybrid IGOA-TLBO algorithm. TLBO is a parameter-free algorithm that simulates the teaching process, including the teacher stage and the learner stage. The core of the hybrid algorithm is to embed the TLBO algorithm into the iteration loop of the IGOA algorithm, and to further optimize the population position by applying the TLBO algorithm after performing the position update of the IGOA in each iteration. In the teacher stage, the population position is updated, and the position update formula is specifically as follows:

[0109]

[0110] In the formula, is the locusti new location, locusts i The original location, rand The random number that follows a uniform distribution within the range (0,1) X Teacher The optimal individual, i.e., the teacher's position. β The teaching factor is randomly assigned a value of 1 or 2. Mean It is the population mean;

[0111] During the learner phase, the population position is updated. The specific formula for position update is as follows:

[0112]

[0113] In the formula, Locusts selected randomly k The original location, The fitness of an individual locust.

[0114] S2 includes the following steps:

[0115] S21. Extract the optimal path node from the path node sequence of the path, and take the starting point as the current point;

[0116] S22. Set the next key point of the current point as the local sub-target point;

[0117] S23. Perform velocity sampling and select the optimal path based on the improved DWA trajectory evaluation function;

[0118] S24. Determine whether the local sub-target point has been reached. If not, return to S23. If yes, proceed to S25.

[0119] S25. Determine if the local sub-target point is the destination. If not, return to S22. If yes, output the planned path from the starting point to the destination.

[0120] In this embodiment, an improved locust optimization algorithm is used for global path planning. After the path is generated, it is used as an intermediate guiding point for local path planning in the DWA algorithm to respond to possible dynamic changes in the environment in real time. The path planning flowchart is as follows. Figure 2 As shown.

[0121] Traditional DWA (Directed Wrap-Up) algorithms only consider factors such as orientation angle, target distance, and velocity to construct an evaluation function, failing to account for the radiation distribution characteristics in the environment. Therefore, they struggle to effectively avoid high-radiation areas in irradiated environments. To address this, a radiation evaluation function, `radiation(v, ω)`, is introduced. This function calculates the average radiation intensity at each sampling point on the candidate trajectory and maps it to a standardized score. A higher score indicates lower radiation exposure and a better path. This improved DWA trajectory evaluation function... The specific expression is as follows:

[0122]

[0123] In the formula, is the radiation evaluation function, is the azimuth evaluation function, is the distance between the robot at the end point of the predicted trajectory and the nearest obstacle on the map, is the linear velocity and angular velocity of the current robot, ([[]] v , ω ) is the velocity space, v is the linear velocity, ω is the angular velocity, , , , and are the weight coefficients of the DWA algorithm;

[0124]

[0125] In the formula, is the average radiation value of each sampling point on the trajectory; [[ID=XX]]

[0126] [[ID=XX]] [[ID=XX]] [[ID=XX]]

[0127] In the formula, is the radiation intensity value of the m sampling point on the trajectory, M is the number of sampling points on the trajectory.

[0128] The evaluation function is a key indicator in path selection, and its weight setting directly affects the effect of path selection. The traditional DWA algorithm usually uses fixed weights to define the evaluation function, which has certain limitations in complex radiation environments because the robot cannot perform adaptive real-time obstacle avoidance according to environmental changes and its own state, and thus is prone to falling into local optimal solutions. Therefore, an adaptive weight strategy is proposed to dynamically adjust the weight coefficients of heading(v, ω) to improve the adaptability of the robot in complex obstacle environments and enhance the obstacle avoidance efficiency. In addition, the present invention also introduces a speed change penalty coefficient to prevent overly sharp turns or speed fluctuations in the path, further enhancing the smoothness and safety of path planning.

[0129] When the obstacle is not near the robot (dist≥P), fixed weights are used to make the robot move towards the target point. When the obstacle reaches near the robot (dist<P), adaptive weights are used to change the weight coefficients of heading(v, ω) according to the distance between the robot and the obstacle. Based on this, in this embodiment, theheading v ω

[0130]

[0131] P is a robot adaptive weight distance threshold value; when the distance between the robot and the obstacle is small, the demand for obstacle avoidance is high, so the weight coefficient of heading(v, ω) is reduced, so that the robot pays more attention to obstacle avoidance and speed adjustment.

[0132] When the linear velocity and angular velocity of the robot change greatly, the weight coefficient of velocity(v, ω) is reduced, so as to suppress the unstable motion. Based on this, the embodiment adjusts the weight coefficients of velocity v ω

[0133]

[0134] ω smooth is a smoothness weight, which will increase with the increase of the speed change rate;

[0135]

[0136] v is the change amount of the linear velocity, ω is the change amount of the angular velocity.

[0137] In the description of the present application, it should be understood that the terms "center", "thickness", "upper", "lower", "horizontal", "top", "bottom", "inner", "outer", "radial" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first", "second", "third" are only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features. Therefore, the features limited by "first", "second", "third" can explicitly or implicitly include one or more of the features.​​​​​​​​​​​​​

Claims

1. A real-time path planning method integrating an improved locust-dynamic window method, characterized in that, Includes the following steps: S1. Generate an initial path based on the improved A* algorithm, and then use the improved locust optimization algorithm to perform global path planning on the path in the radiation environment to determine the path from the starting point to the ending point. S2. The improved DWA algorithm is used to perform local path planning based on the path to generate a planned path; S1 includes the following steps: S11. Use the improved A* algorithm to generate the initial path and obtain the initial position of the population; Calculate the fitness value of each locust individual in the population, select the optimal fitness value as the target fitness value; initialize the number of locusts and the current iteration number, and set the maximum number of iterations; S12. Set the number of locusts to 1; S13. Update the population location based on the location update strategy that takes into account radiation distribution information; S14. Calculate the current fitness value based on the current population position, and determine whether the current fitness value is better than the target fitness value; if yes, update the target fitness value and the target optimal solution position based on the current fitness value and the current population position, and proceed to S15; if no, proceed to S15. S15. Determine whether the set population size has been reached. If yes, proceed to S16. If no, increment the current number of locusts by one and return to S13. S16. Update the population position according to the blended learning optimization algorithm; S17. Determine if the current iteration count has reached the maximum iteration count. If yes, output the position of the target optimal solution and generate a path from the starting point to the ending point. If no, increment the current iteration count and return to S12. In S11, the improved A* algorithm is optimized by introducing a multi-scale evaluation method. H ( n l The estimation process, optimization H ( n l The specific expression for ) is: In the formula, λ As a weighting factor, L i The distance between two adjacent points. For the first a Local radiation dose at +1 node, For the first a Radiation dose rate at each node For the first a Radiation dose rate of +1 node v t For movement speed, N This represents the number of points between the current point and the destination. In the formula, For the first a +1 node surrounding the first b Radiation dose at adjacent points For the first b The weight of each point; S2 includes the following steps: S21. Extract the optimal path node from the path node sequence of the path, and take the starting point as the current point; S22. Set the next key point of the current point as the local sub-target point; S23. Perform velocity sampling and select the optimal path based on the improved DWA trajectory evaluation function; S24. Determine whether the local sub-target point has been reached. If not, return to S23. If yes, proceed to S25. S25. Determine if the local sub-target point is the destination. If not, return to S22. If yes, output the planned path from the starting point to the destination. In S23, the improved DWA trajectory evaluation function The specific expression is: In the formula, For radiation evaluation function, For azimuth evaluation function, This is the distance between the robot and the nearest obstacle on the map when the robot is at the end point of the predicted trajectory. Given the current linear velocity and angular velocity of the robot, v , ω ( ) represents the velocity space. v For linear velocity, ω Angular velocity, , , , and These are the weight coefficients for the DWA algorithm; In the formula, This represents the average radiation value at each sampling point on the trajectory; In the formula, For the first on the trajectory m The radiation intensity value at the sampling point M This represents the number of sampling points on the trajectory.

2. The real-time path planning method integrating the improved locust-dynamic window method according to claim 1, characterized in that, In S13, the location update strategy considering radiation distribution information is used to calculate locusts. i Location The specific expression is: In the formula, b max This is the upper bound of the search space. b min This serves as the lower bound of the search space. T d The optimal value among the targets. d ij locusts i and j The distance between them locusts i Location to locusts j The unit vector at the location, For gradient weights, For the current position in dimension d The radiation gradient on the surface is estimated using the finite difference method. N For population size, c For adaptive comfort zone coefficient, n L is the current iteration number, and L is the maximum iteration number. The function used to calculate the interaction forces among locust populations is specifically expressed as follows: In the formula, f For the attraction intensity parameter, l For the attraction scale parameter.

3. The real-time path planning method integrating the improved locust-dynamic window method according to claim 2, characterized in that, In S13, the adaptive comfort zone coefficient c The expression for adaptive adjustment based on radiation intensity is as follows: In the formula, c max This represents the maximum value of the adaptive comfort zone coefficient. c min This represents the minimum adaptive comfort zone coefficient. α As a regulating factor, For the first i Only the locust's current location is radiating. R min This represents the minimum radiation value on the map. R max This represents the maximum radiation value on the map.

4. The real-time path planning method integrating the improved locust-dynamic window method according to claim 1, characterized in that, S16 includes the following steps: S161. Update the population position during the teacher stage. The specific position update formula is as follows: In the formula, locusts i new location, locusts i The original location, rand The random number that follows a uniform distribution within the range (0,1) X Teacher For the optimal individual, β As a teaching factor, Mean It is the population mean; S162. Update the population position during the learner phase. The specific formula for position update is as follows: In the formula, Locusts selected randomly k The original location, The fitness of an individual locust.

5. The real-time path planning method integrating the improved locust-dynamic window method according to claim 1, characterized in that, Adjust according to the distance between the robot and the obstacle. heading ( v , ω Weighting coefficients Adjusted weighting coefficients The specific expression is: In the formula, P For robot adaptive weight distance threshold; Adjust according to the robot's linear velocity and angular velocity velocity ( v , ω Weighting coefficients Adjusted weighting coefficients The specific expression is: In the formula, ω smooth For smoothness weights; In the formula, v Δ is the change in linear velocity. ω Δ represents the change in angular velocity.

Images