Accurate positioning and path planning method for nuclear power robot

Abstract

The invention relates to the technical field of nuclear power robots, discloses a nuclear power robot accurate positioning and path planning method, and aims to construct a three-dimensional environment model for a complex and changeable environment in a nuclear power station, and express the environment as a grid map containing environment characteristic information. According to the method, a comprehensive state space is defined, the three-dimensional position, obstacle distribution, radiation intensity and energy state of the nuclear power robot are covered, and real-time updating of dynamic obstacles is considered. The action space defines all possible moving actions of the robot so as to ensure the planning flexibility. By defining a comprehensive cost function, the length of a path, radiation exposure, energy consumption and collision risk are evaluated, and a strategy is formulated with the purpose of minimizing accumulated cost. According to the method, the heuristic search algorithm is combined with the reinforcement learning strategy to carry out path planning, so that the planning precision and adaptability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of nuclear power robots, and in particular to a method for precise positioning and path planning of nuclear power robots. Background Art

[0002] In the operation and maintenance of nuclear power plants, nuclear power robots are increasingly used. They undertake various tasks such as inspection, testing, and maintenance to replace manual work in extreme environments such as high radiation, high temperature, and high pressure. However, the internal environment of a nuclear power plant is complex and changeable, with a large number of static and dynamic obstacles and unevenly distributed radiation fields, which places extremely high demands on the positioning and path planning of nuclear power robots.

[0003] Current path planning methods are often based on two-dimensional plane models, ignoring the three-dimensional characteristics of the internal environment of nuclear power plants, such as the three-dimensional layout of equipment, platforms at different heights, etc., resulting in the planned paths being impossible to execute or inefficient in actual applications. At the same time, these methods usually only consider path length as the optimization target, while ignoring key factors such as radiation exposure, energy consumption, and collision risk, which may cause nuclear power robots to face higher risks and costs when performing tasks.

[0004] In addition, the dynamic nature of the nuclear power plant's internal environment also poses challenges to path planning. The appearance and movement of dynamic obstacles, such as other mobile devices or human activities, require the path planning algorithm to be able to update and replan the path in real time to ensure the safe and efficient operation of the nuclear power robot. Traditional path planning methods often lack sufficient flexibility and adaptability to cope with such a dynamically changing environment. Summary of the invention

[0005] The purpose of the present invention is to provide a nuclear power robot precise positioning and path planning method to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a nuclear power robot precise positioning and path planning method, the method comprising:

[0007] Construct a three-dimensional environment model for the nuclear power robot and represent the internal environment of the nuclear power plant as a grid map, where each grid contains environmental feature information;

[0008] The state space is defined, including the current position, target position, surrounding obstacle distribution, radiation intensity distribution and robot energy state of the nuclear power robot; the current position and target position are represented by three-dimensional coordinates; the surrounding obstacle distribution includes static obstacles and dynamic obstacles, the static obstacles are fixed structures inside the nuclear power plant, and the dynamic obstacles are other mobile devices or personnel; the radiation intensity distribution is obtained through pre-measured radiation data and updated in real time; the robot energy state includes battery power and working time;

[0009] Define the action space, which includes the possible actions that the nuclear power robot may take during movement, including forward, backward, left turn, right turn, ascend, descend, accelerate and decelerate;

[0010] Define a cost function to evaluate the path planning results based on path length, radiation exposure, energy consumption, and collision risk, and calculate the corresponding cost;

[0011] Formulate strategies, rules for selecting actions in a given state, and continuously iterate through optimization algorithms to minimize cumulative costs;

[0012] A heuristic search algorithm combined with a reinforcement learning strategy is used to plan the path of the nuclear power robot.

[0013] Preferably, the path length is calculated based on the actual moving distance of the nuclear power robot from the current position to the target position; the radiation exposure is calculated by cumulatively adding the radiation intensity of each grid passed by the nuclear power robot during the movement; the energy consumption is calculated based on the energy consumption model of the nuclear power robot when performing the action; and the collision risk is evaluated based on the distance and relative speed between the nuclear power robot and the surrounding obstacles.

[0014] Preferably, the current position data, surrounding obstacle distribution data and radiation intensity data are collected in real time through sensors and navigation systems installed on the nuclear power robot as raw data of the state space, and the environmental feature information in the grid map is updated.

[0015] Preferably, the components of the cost function include path length cost, radiation exposure cost, energy consumption cost and collision risk cost:

[0016] The calculation formula of the path length cost function is: C_length = k1*d, where d is the path length and k1 is the path length cost coefficient;

[0017] The calculation formula of the radiation exposure cost function is: C_radiation = k2*Σr_i, where Σr_i is the cumulative value of the radiation intensity of each grid on the path, and k2 is the radiation exposure cost coefficient;

[0018] The calculation formula of the energy consumption cost function is: C_energy = k3*E, where E is the total energy consumption of the robot on the path, and k3 is the energy consumption cost coefficient;

[0019] The calculation formula of the collision risk cost function is: C_collision = k4*max(0,1-d_safe / d_min), where d_safe is the safety distance threshold, d_min is the distance between the robot and the nearest obstacle, and k4 is the collision risk cost coefficient;

[0020] Through the above cost functions, the final cost function C is obtained: C=C_length+C_radiation+C_energy+C_collision.

[0021] Preferably, a heuristic search algorithm combined with a Q-learning algorithm is used for path planning, wherein the heuristic search algorithm is used to preliminarily generate a feasible path, and the Q-learning algorithm is used to optimize the path.

[0022] Preferably, the steps of optimizing the path by combining the heuristic search algorithm with the Q-learning algorithm include:

[0023] S1: Initialize the Q value table, set the learning rate α, discount factor γ and exploration rate ε;

[0024] S2: Generate a preliminary path from the current location to the target location using a heuristic search algorithm;

[0025] S3: The nuclear power robot moves along the preliminary path, and each time it reaches a new state, it updates the Q value in the Q value table according to the current state, the selected action, and the instant cost obtained;

[0026] S4: During the movement process, the action is selected according to the ε-greedy strategy, that is, a random action is selected with a probability of ε, and the action with the largest current Q value is selected with a probability of 1-ε;

[0027] S5: Repeat steps S3 and S4 until the nuclear power robot reaches the target position or reaches a preset number of learning rounds;

[0028] S6: Based on the learned Q-value table, a heuristic search algorithm is used to regenerate the optimized path as the final path planning result.

[0029] Preferably, the implementation steps of the heuristic search algorithm include:

[0030] A1: Initialize the open list and the closed list to be empty; take the starting position of the nuclear power robot as the initial node, add it to the open list, set the parent node of the node to be empty, the path cost to 0, and estimate the total cost as the heuristic cost from the starting position to the target position;

[0031] A2: Check whether the open list is empty. If it is empty, the path does not exist and the algorithm ends. Otherwise, take the node with the smallest estimated total cost from the open list as the current node, remove it from the open list, and add it to the closed list.

[0032] A3: Determine whether the current node is the target position. If so, build a path and the algorithm ends. The steps of building a path include: starting from the current node, tracing back along the parent node to the starting position to form a path;

[0033] A4: Get all the adjacent nodes of the current node. For each adjacent node, perform the following operations:

[0034] If the adjacent node is in the closed list, skip the node;

[0035] If the adjacent node is not in the open list, calculate the path cost of the adjacent node, which is the path cost of the current node plus the movement cost from the current node to the adjacent node; calculate the estimated total cost of the adjacent node, which is the path cost of the adjacent node plus the heuristic cost from the adjacent node to the target location; add the adjacent node to the open list and set its parent node to the current node;

[0036] If the adjacent node is already in the open list, and the path cost from the current node to the adjacent node is less than the original path cost, then update the path cost of the adjacent node and the parent node to the current node, and update the estimated total cost of the adjacent node;

[0037] A5: Repeat steps A2 to A4 until the target location is found or the open list is empty.

[0038] Preferably, the step of dynamically adjusting the heuristic cost includes: monitoring real-time data of the internal environment of the nuclear power plant, including radiation intensity, obstacle location and robot energy status; calculating the weight coefficient of each influencing factor based on the real-time data, wherein the higher the radiation intensity, the denser the obstacles or the lower the robot energy, the greater the weight coefficient of the corresponding influencing factor; applying the weight coefficient to the calculation of the heuristic cost, so that the heuristic cost is closer to the actual situation.

[0039] Preferably, the optimization management strategy for the open list and the closed list is implemented by adopting a priority queue data structure to store the open list and sorting the nodes according to the estimated total cost.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] By constructing a three-dimensional environment model of the nuclear power robot and representing the internal environment of the nuclear power plant as a grid map, the present invention fully considers the three-dimensional characteristics of the environment, such as the three-dimensional layout and platforms at different heights. This makes the planned path closer to reality, improves the feasibility and efficiency of the path, and avoids the limitations brought by the traditional two-dimensional plane model.

[0042] The present invention defines a comprehensive state space including the current position of the nuclear power robot, the target position, the distribution of obstacles in the surrounding environment, the distribution of radiation intensity, and the energy state of the robot. This comprehensive state description enables the path planning algorithm to more accurately assess the environment, thereby planning a safer and more efficient path.

[0043] By defining a cost function including path length, radiation exposure, energy consumption and collision risk, the present invention achieves a multi-dimensional evaluation of path planning results. This helps to comprehensively consider various key factors when planning the path and reduce the risk and cost of the nuclear power robot when performing tasks.

[0044] The present invention formulates rules for selecting actions in a given state and continuously iterates through optimization algorithms to minimize the cumulative cost. This strategy formulation method enables the path planning algorithm to be flexibly adjusted according to real-time environmental information, improving the adaptability and robustness of the algorithm.

[0045] By adopting a heuristic search algorithm combined with a reinforcement learning strategy, the present invention realizes efficient path planning for a nuclear power robot. The heuristic search algorithm can quickly find an approximate optimal solution, while the reinforcement learning strategy can further improve the quality and efficiency of path planning through continuous learning and optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a step diagram of the nuclear power robot precise positioning and path planning method of the present invention;

[0047] Figure 2 It is a schematic diagram of the design of the cost function;

[0048] Figure 3 This is a schematic diagram of the path planning principle of the present invention combining the heuristic search algorithm and the Q-learning algorithm. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0050] See also Figure 1-3 The present invention provides a technical solution: a method for precise positioning and path planning of a nuclear power robot, the method comprising:

[0051] Construct a three-dimensional environment model for the nuclear power robot. The internal environment of the nuclear power plant is represented in the form of a grid map, where each grid contains detailed environmental feature information, such as terrain type, obstacle location, radiation intensity, etc. This three-dimensional environment model can fully reflect the complexity and three-dimensionality of the internal environment of the nuclear power plant, and provide accurate basic data for subsequent path planning.

[0052] Define the state space. The state space covers multiple dimensions such as the current position of the nuclear power robot, the target position, the distribution of obstacles in the surrounding environment, the distribution of radiation intensity, and the energy state of the robot. Among them, the current position and the target position are represented by precise three-dimensional coordinates to ensure the accuracy of path planning. The distribution of obstacles in the surrounding environment includes static obstacles (such as fixed structures inside the nuclear power plant) and dynamic obstacles (such as other mobile devices or personnel) to achieve a comprehensive perception of the dynamic environment. The radiation intensity distribution is obtained through pre-measured radiation data and is updated in real time according to actual conditions to ensure that the impact of radiation exposure can be fully considered during the path planning process. The energy state of the robot includes battery power and working time, which provides energy constraints for path planning.

[0053] Define the action space. The action space contains all possible actions that the nuclear power robot may take during movement, such as forward, backward, left turn, right turn, ascend, descend, accelerate and decelerate, etc. The definition of these actions provides a rich selection space for path planning, allowing the nuclear power robot to flexibly adjust its movement strategy according to actual needs.

[0054] Define the cost function. The cost function is a key evaluation indicator in the path planning process. It takes into account multiple factors such as path length, radiation exposure, energy consumption, and collision risk. By assigning reasonable weights to these factors, the total cost of each path can be calculated, providing a basis for subsequent strategy formulation.

[0055] When formulating the strategy, the present invention adopts a rule for selecting actions under a given state. The optimization algorithm is continuously iterated to minimize the cumulative cost of the nuclear power robot in the process of performing the task. This strategy can ensure that the nuclear power robot completes the path planning in the best way while meeting the task requirements.

[0056] Finally, the present invention adopts a heuristic search algorithm combined with a reinforcement learning strategy to plan the path of the nuclear power robot. The heuristic search algorithm can quickly find the approximate optimal solution, while the reinforcement learning strategy can enable the nuclear power robot to gradually master the ability to plan paths in complex environments through continuous trial and error and learning. This combination not only improves the efficiency of path planning, but also enhances the adaptability and robustness of the nuclear power robot to dynamic environments.

[0057] This outbreak is further described below in conjunction with Examples 1 to 3:

[0058] Embodiment 1:

[0059] This embodiment is used to describe the calculation method of path length, radiation exposure, energy consumption and collision risk and the construction process of the cost function, and also explains how to update the state space data in real time through sensors and navigation systems.

[0060] First, for the calculation of path length, the present invention uses the actual moving distance as a measurement standard. During the movement of the nuclear power robot, the actual moving distance d from the current position to the target position can be accurately calculated through the trajectory data recorded by the navigation system. The path length cost function C_length is calculated according to the formula C_length = k1*d, where k1 is the path length cost coefficient, which is used to adjust the proportion of path length in the total cost.

[0061] The calculation of radiation exposure takes into account the radiation intensity of each grid that the nuclear power robot passes through during movement. The radiation sensor installed on the nuclear power robot can measure and record the radiation intensity value of each grid that the robot passes through in real time. The radiation exposure cost function C_radiation is calculated according to the formula C_radiation = k2*Σr_i, where Σr_i is the cumulative value of the radiation intensity of each grid on the path, and k2 is the radiation exposure cost coefficient, which is used to reflect the degree of influence of radiation exposure on the total cost.

[0062] In terms of the calculation of energy consumption, the present invention is based on the energy consumption model of the nuclear power robot when performing actions. The energy consumption model takes into account the energy consumed by different types of robot actions (such as forward, backward, left turn, right turn, etc.). By real-time monitoring of the robot's action execution and combining the energy consumption model, the total energy consumption E of the robot on the path can be calculated. The energy consumption cost function C_energy is calculated according to the formula C_energy=k3*E, where k3 is the energy consumption cost coefficient, which is used to adjust the proportion of energy consumption in the total cost.

[0063] For the assessment of collision risk, the present invention takes into account the distance and relative speed between the nuclear power robot and the surrounding obstacles. Through the sensors installed on the nuclear power robot, the distance d_min between the robot and the surrounding obstacles can be monitored in real time. At the same time, the degree of collision risk can be evaluated by combining the moving speed of the obstacle and the moving speed of the robot. The collision risk cost function C_collision is calculated according to the formula C_collision=k4*max(0,1-d_safe / d_min), where d_safe is the preset safety distance threshold, and k4 is the collision risk cost coefficient, which is used to reflect the degree of impact of the collision risk on the total cost. When d_min is less than d_safe, the collision risk cost increases; when d_min is greater than or equal to d_safe, the collision risk cost is 0.

[0064] In order to update the state space data in real time, the present invention uses the sensors and navigation system installed on the nuclear power robot to collect the current position data, the surrounding obstacle distribution data and the radiation intensity data in real time. These data are used as the original data of the state space to update the environmental feature information in the grid map. By continuously updating the grid map, it can be ensured that the data used in the path planning process is the latest and accurate.

[0065] Embodiment 2:

[0066] This embodiment is used to describe a method for nuclear power robot path planning by combining a heuristic search algorithm with a Q-learning algorithm, aiming to achieve efficient and safe navigation of the nuclear power robot in a complex environment by preliminarily generating a feasible path and optimizing it. The following are the detailed implementation steps of the method:

[0067] S1: Initialize Q value table and parameter settings

[0068] A Q-value table of state-action pairs is constructed for the nuclear power robot, which is used to store the expected cumulative cost of taking different actions in each state. At the same time, the learning rate α (used to control the update step size in the learning process), discount factor γ (used to calculate the impact of future costs on current decisions) and exploration rate ε (used to balance the relationship between exploration and utilization) of the Q-learning algorithm are set. The values ​​of these parameters need to be adjusted according to the actual application scenario and experimental results.

[0069] S2: Heuristic search algorithm generates preliminary paths

[0070] A heuristic search algorithm (such as A* algorithm, Dijkstra algorithm, etc.) is used to generate a preliminary path from the current position of the nuclear power robot to the target position based on the grid map of the internal environment of the nuclear power plant and the preset heuristic function (such as the weighted sum of path length, radiation exposure, etc.). This path serves as the basis for subsequent Q-learning algorithm learning and optimization.

[0071] S3: Nuclear power robot movement and Q value update

[0072] The nuclear power robot moves along the preliminary path, and each time it reaches a new state (i.e. a node in the grid map), it performs the following operations:

[0073] Observe the current state and select an action based on the Q value table. At this time, according to the ε-greedy strategy, an action is randomly selected with a probability of ε, and the action with the largest current Q value is selected with a probability of 1-ε to balance the needs of exploring new paths and using known optimal paths.

[0074] Execute the selected action and observe the resulting immediate cost (including path length cost, radiation exposure cost, energy consumption cost, collision risk cost, etc.).

[0075] According to the update rules of the Q-learning algorithm, the Q value of the corresponding state-action pair in the Q value table is updated using the immediate cost, current Q value, learning rate α and discount factor γ.

[0076] S4: ε-greedy strategy selects actions

[0077] During the movement of the nuclear power robot, the ε-greedy strategy is continuously applied to select actions. As the learning process proceeds, the value of ε can be gradually reduced to increase the probability of using the known optimal path and reduce meaningless exploration.

[0078] S5: Repeated learning and exploration

[0079] Repeat steps S3 and S4 until the nuclear power robot successfully reaches the target position or reaches the preset number of learning rounds. During the learning process, the nuclear power robot will continuously accumulate experience about different state-action pairs and reflect these experiences by updating the Q value table.

[0080] S6: Regenerate optimized path

[0081] After learning is completed, the heuristic search algorithm (which may be the same or different from the algorithm in step S2) is used again to regenerate the path based on the learned Q value table. At this time, the heuristic function can rely more on the information in the Q value table to more accurately evaluate the pros and cons of different paths. The generated path is used as the final path planning result to guide the actual movement of the nuclear power robot.

[0082] Through the above steps, the present invention realizes the effective combination of the heuristic search algorithm and the Q-learning algorithm, which not only utilizes the ability of the heuristic search algorithm to quickly generate a preliminary path, but also improves the accuracy and adaptability of the path planning results through the learning and optimization of the Q-learning algorithm.

[0083] Embodiment 3:

[0084] The heuristic search algorithm is used to preliminarily generate feasible paths. The following is the detailed implementation of the heuristic search algorithm and its optimization strategy:

[0085] Heuristic search algorithm implementation steps

[0086] A1. Initialization phase:

[0087] Initialize the open list (used to store nodes to be explored) and the closed list (used to store nodes that have been explored), both of which are empty.

[0088] The starting position of the nuclear power robot is taken as the initial node and added to the open list.

[0089] Set the parent node of the initial node to empty (indicating that there is no predecessor node), the path cost to 0 (indicating that the actual moving cost from the starting position to the node is 0), and the estimated total cost is the heuristic cost from the starting position to the target position (which can be estimated based on factors such as distance, radiation intensity, and energy consumption).

[0090] A2. Node selection and expansion:

[0091] Check if the open list is empty. If it is empty, it means there is no feasible path and the algorithm ends.

[0092] The node with the smallest estimated total cost is taken from the open list as the current node. This is usually implemented through a priority queue data structure to ensure that the node with the smallest estimated total cost can be found quickly every time.

[0093] Remove the current node from the open list and add it to the closed list, indicating that the node has been explored.

[0094] A3. Target determination and path construction:

[0095] Determine whether the current node is the target position. If so, execute the path construction step: starting from the current node, trace back along the parent node to the starting position to form a path from the starting position to the target position, and the algorithm ends.

[0096] A4. Adjacent node processing:

[0097] Get all the neighboring nodes of the current node and do the following for each neighboring node:

[0098] If the adjacent node is already in the closed list, it means that the node has been explored and the node is skipped.

[0099] If the adjacent node is not in the open list, calculate its path cost (the path cost of the current node plus the movement cost from the current node to the adjacent node), estimate the total cost (path cost plus the heuristic cost from the adjacent node to the target location), add it to the open list, and set its parent node to the current node.

[0100] If the adjacent node is already in the open list, but the path cost to reach the adjacent node through the current node is less than the original path cost, update the path cost, parent node and estimated total cost of the adjacent node.

[0101] A5. Repeated iteration:

[0102] Repeat steps A2 to A4 until the target location is found or the open list is empty (indicating no feasible path).

[0103] Monitor the real-time data of the internal environment of the nuclear power plant, including radiation intensity, obstacle location, and robot energy status. According to the real-time data, calculate the weight coefficient of each influencing factor. For example, the higher the radiation intensity, the greater its weight coefficient; the denser the obstacles, the greater its corresponding weight; the lower the robot energy, the greater the weight of its energy status. Applying these weight coefficients to the calculation of heuristic cost makes the heuristic cost closer to the actual situation and improves the accuracy and practicality of path planning.

[0104] The open list is stored in a priority queue data structure, and the nodes are sorted according to the estimated total cost. In this way, each time a node is selected, the node with the smallest estimated total cost can be quickly found, which improves the efficiency of the algorithm. For the closed list, a data structure such as a hash table or set can be used to store it, so as to quickly determine whether a node has been explored and avoid repeated exploration.

[0105] 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.

[0106] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A nuclear power robot precise positioning and path planning method, characterized in that: The method comprises: Construct a three-dimensional environment model for the nuclear power robot and represent the internal environment of the nuclear power plant as a grid map, where each grid contains environmental feature information; The state space is defined, including the current position, target position, surrounding obstacle distribution, radiation intensity distribution and robot energy state of the nuclear power robot; the current position and target position are represented by three-dimensional coordinates; the surrounding obstacle distribution includes static obstacles and dynamic obstacles, the static obstacles are fixed structures inside the nuclear power plant, and the dynamic obstacles are other mobile devices or personnel; the radiation intensity distribution is obtained through pre-measured radiation data and updated in real time; the robot energy state includes battery power and working time; Define the action space, which includes the possible actions that the nuclear power robot may take during movement, including forward, backward, left turn, right turn, ascend, descend, accelerate and decelerate; Define a cost function to evaluate the path planning results based on path length, radiation exposure, energy consumption, and collision risk, and calculate the corresponding cost; Formulate strategies, rules for selecting actions in a given state, and continuously iterate through optimization algorithms to minimize cumulative costs; A heuristic search algorithm combined with a reinforcement learning strategy is used to plan the path of the nuclear power robot.

2. The nuclear power robot precise positioning and path planning method according to claim 1 is characterized by: The path length is calculated based on the actual moving distance of the nuclear power robot from the current position to the target position; the radiation exposure is calculated based on the radiation intensity of each grid that the nuclear power robot passes through during the movement; The energy consumption is calculated based on the energy consumption model of the nuclear power robot when performing actions; the collision risk is evaluated based on the distance and relative speed between the nuclear power robot and surrounding obstacles.

3. The nuclear power robot precise positioning and path planning method according to claim 2 is characterized in that: Through the sensors and navigation system installed on the nuclear power robot, the current position data, surrounding obstacle distribution data and radiation intensity data are collected in real time as the original data of the state space, and the environmental feature information in the grid map is updated.

4. The nuclear power robot precise positioning and path planning method according to claim 3 is characterized in that: The components of the cost function include path length cost, radiation exposure cost, energy consumption cost, and collision risk cost: The calculation formula of the path length cost function is: C_length = k1*d, where d is the path length and k1 is the path length cost coefficient; The calculation formula of the radiation exposure cost function is: C_radiation = k2*Σr_i, where Σr_i is the cumulative value of the radiation intensity of each grid on the path, and k2 is the radiation exposure cost coefficient; The calculation formula of the energy consumption cost function is: C_energy = k3*E, where E is the total energy consumption of the robot on the path, and k3 is the energy consumption cost coefficient; The calculation formula of the collision risk cost function is: C_collision = k4*max(0,1-d_safe / d_min), where d_safe is the safety distance threshold, d_min is the distance between the robot and the nearest obstacle, and k4 is the collision risk cost coefficient; Through the above cost functions, the final cost function C is obtained: C=C_length+C_radiation+C_energy+C_collision.

5. The nuclear power robot precise positioning and path planning method according to claim 4 is characterized in that: A heuristic search algorithm combined with a Q-learning algorithm is used for path planning, where the heuristic search algorithm is used to preliminarily generate feasible paths, and the Q-learning algorithm is used to optimize the path.

6. The nuclear power robot precise positioning and path planning method according to claim 5 is characterized in that: The steps of combining the heuristic search algorithm with the Q-learning algorithm to optimize the path include: S1: Initialize the Q value table, set the learning rate α, discount factor γ and exploration rate ε; S2: Generate a preliminary path from the current location to the target location using a heuristic search algorithm; S3: The nuclear power robot moves along the preliminary path, and each time it reaches a new state, it updates the Q value in the Q value table according to the current state, the selected action, and the instant cost obtained; S4: During the movement process, the action is selected according to the ε-greedy strategy, that is, a random action is selected with a probability of ε, and the action with the largest current Q value is selected with a probability of 1-ε; S5: Repeat steps S3 and S4 until the nuclear power robot reaches the target position or reaches a preset number of learning rounds; S6: Based on the learned Q-value table, a heuristic search algorithm is used to regenerate the optimized path as the final path planning result.

7. The nuclear power robot precise positioning and path planning method according to claim 6 is characterized in that: The implementation steps of the heuristic search algorithm include: A1: Initialize the open list and the closed list to be empty; take the starting position of the nuclear power robot as the initial node, add it to the open list, set the parent node of the node to be empty, the path cost to 0, and estimate the total cost as the heuristic cost from the starting position to the target position; A2: Check whether the open list is empty. If it is empty, the path does not exist and the algorithm ends. Otherwise, take the node with the smallest estimated total cost from the open list as the current node, remove it from the open list, and add it to the closed list. A3: Determine whether the current node is the target position. If so, build a path and the algorithm ends. The steps of building a path include: starting from the current node, tracing back along the parent node to the starting position to form a path; A4: Get all the adjacent nodes of the current node. For each adjacent node, perform the following operations: If the adjacent node is in the closed list, skip the node; If the adjacent node is not in the open list, calculate the path cost of the adjacent node, which is the path cost of the current node plus the movement cost from the current node to the adjacent node; calculate the estimated total cost of the adjacent node, which is the path cost of the adjacent node plus the heuristic cost from the adjacent node to the target location; add the adjacent node to the open list and set its parent node to the current node; If the adjacent node is already in the open list, and the path cost from the current node to the adjacent node is less than the original path cost, then update the path cost of the adjacent node and the parent node to the current node, and update the estimated total cost of the adjacent node; A5: Repeat steps A2 to A4 until the target location is found or the open list is empty.

8. The nuclear power robot precise positioning and path planning method according to claim 7 is characterized in that: The steps of dynamically adjusting the heuristic cost include: monitoring real-time data of the internal environment of the nuclear power plant, including radiation intensity, obstacle positions and robot energy status; calculating the weight coefficient of each influencing factor based on the real-time data, wherein the higher the radiation intensity, the denser the obstacles or the lower the robot energy, the greater the weight coefficient of the corresponding influencing factor; applying the weight coefficient to the calculation of the heuristic cost, so that the heuristic cost is closer to the actual situation.

9. The nuclear power robot precise positioning and path planning method according to claim 8, characterized in that: An optimized management strategy for the open list and the closed list is proposed by adopting a priority queue data structure to store the open list and sorting the nodes according to the estimated total cost.

Images