Navigation method of intelligent navigation robot based on multi-sensor fusion

Through multi-sensor fusion and advanced path planning algorithms, the navigation accuracy and security problems of robot navigation in complex environments are solved, and efficient and adaptive navigation capabilities are achieved.

CN118936472APending Publication Date: 2024-11-12NAVAL UNIV OF ENG PLA

Patent Information

Application Number
CN202410990494.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-23
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The existing robot navigation technology is insufficient in complex environments, with low data reliability of a single sensor, slow response speed in dynamic environments, and lack of adaptability, resulting in reduced navigation accuracy and safety impacts.

Method used

Multi-sensor fusion technology is adopted, combined with the dung beetle optimization algorithm and the Dijkstra algorithm, and environmental data is obtained through lidar, vision sensor and ultrasonic sensor, a three-dimensional environmental model is constructed, and a particle filtering algorithm is used to perform real-time position correction and dynamic obstacle avoidance.

Benefits of technology

It improves the accuracy and reliability of the environment, realizes high-precision navigation and adaptive obstacle avoidance in complex environments, and significantly improves the safety of the robot and task completion efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118936472A_ABST
    Figure CN118936472A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of dung beetle optimization methods, discloses a robot path planning system and method based on a dung beetle optimization algorithm, and realizes comprehensive perception and accurate modeling of a complex three-dimensional environment by introducing multi-sensor fusion and an advanced environment modeling technology. The robot can dynamically update the environment model according to environment data collected in real time, and sensitivity and adaptability to environment changes are kept. Meanwhile, a self-adaptive navigation algorithm is adopted, so that the robot can autonomously adjust a navigation strategy and path planning according to task requirements and environment changes. The robot can flexibly cope with various challenges in complex and dynamic environments due to the high self-adaptive capacity, and the reliability and the intelligent level of the robot in practical application are remarkably improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent robots, and in particular relates to a navigation method of an intelligent navigation robot based on multi-sensor fusion. Background Art

[0002] The following are two existing robot path planning methods and their problems:

[0003] Method 1: Path planning based on Dijkstra algorithm

[0004] Dijkstra's algorithm is a classic shortest path algorithm used to calculate the shortest path in a static environment. In robot path planning, this algorithm can be used to determine the optimal or suboptimal path of a robot in a known environment.

[0005] Problems:

[0006] 1. Environmental assumptions: Dijkstra's algorithm usually assumes that the environment is static and does not consider the movement or changes of obstacles. However, in practical applications, robots may encounter dynamically changing environments, which may cause the path planned by the algorithm to become invalid.

[0007] 2. Computational resource consumption: Although the Dijkstra algorithm performs well in static environments, its computational complexity may be high in large-scale or complex environments, requiring a large amount of computing resources.

[0008] Method 2: RRT (Rapidly-exploring Random Trees) algorithm

[0009] The RRT algorithm is an algorithm that builds a search tree based on random sampling and selecting the nearest neighbor nodes. It is suitable for robot path planning in non-convex dynamic environments and can quickly find paths in large-scale spaces.

[0010] Problems:

[0011] 1. Path optimization problem: Although the RRT algorithm can quickly find a feasible path, the generated path is often not optimal and may require further post-processing to optimize the path length, smoothness, etc.

[0012] 2. Dynamic environment adaptability: Similar to the Dijkstra algorithm, the RRT algorithm also faces the challenge of dynamic environment adaptability. When obstacles move in the environment, it may be necessary to rerun the algorithm to find a new path.

[0013] Existing robot path planning methods have their own advantages and disadvantages, and it is necessary to select appropriate algorithms based on specific application scenarios and requirements. At the same time, with the continuous development of robotics technology, higher requirements are also placed on the performance and adaptability of algorithms.

[0014] The Dung Beetle Optimizer (DBO) algorithm is a new swarm intelligence algorithm proposed in 2022. The algorithm performs different search methods based on subpopulations and adopts a dynamic boundary search strategy to reduce the search space. Compared with other algorithms, it has strong optimization ability and fast convergence speed. At present, the algorithm has been widely used in different fields. However, because the dung beetle is limited by the ability to transmit information during the search process, the algorithm's global exploration and local development capabilities are unbalanced, it is easy to fall into local optimality, and the global search ability is weak, which will affect the application of DBO in practical problems.

[0015] The Dung Beetle Optimizer (DBO) algorithm may face the following technical problems and defects in robotic applications:

[0016] 1) Balance between global search and local search:

[0017] When dealing with complex optimization problems, the DBO algorithm may fall into a local optimal solution due to the imbalance between global exploration and local development capabilities. In robot applications, such as path planning and target tracking, if the algorithm cannot effectively balance global search and local fine search, the robot may miss a better solution, thus affecting the robot's performance.

[0018] 2) Limitation of information transmission capabilities:

[0019] The search behavior of dung beetles in nature is limited by their ability to transmit information, which leads to similar problems when algorithms simulate this behavior. In robotic applications, if the algorithm cannot effectively transmit information, such as environmental information and task goals, the robot's decision-making efficiency and accuracy may be affected.

[0020] 3) Dynamic environment adaptability:

[0021] When a robot works in a real environment, it may encounter dynamically changing environments, such as moving obstacles, changing lighting, etc. When dealing with such dynamic problems, the DBO algorithm may not be able to quickly adapt to changes in the environment due to its fixed search strategy and boundary adjustment mechanism, resulting in the robot being unable to effectively complete the task.

[0022] 4) Sensitivity of parameter settings:

[0023] The performance of the DBO algorithm is often affected by parameter settings. In robot applications, improper parameter settings may lead to problems such as slow algorithm convergence and low search accuracy, which in turn affects the working effect of the robot.

[0024] 5) Multi-objective optimization problem:

[0025] In robotic applications, it is often necessary to optimize multiple objectives at the same time, such as path length, time, energy consumption, etc. When dealing with such multi-objective optimization problems, the DBO algorithm may need to be combined with other algorithms or the algorithm may need to be improved to better meet the needs of multi-objective optimization.

[0026] Through the above analysis, the problems and defects of the existing technology are as follows: the existing robot navigation technology still has many shortcomings in its performance in complex environments. First, the reliability of single sensor data is low, and it is easily disturbed by environmental noise and obstacles, resulting in reduced navigation accuracy. Secondly, the existing path planning algorithm has a slow response speed in a dynamic environment and cannot avoid obstacles in real time, which affects the safety of the robot and the efficiency of task completion. In addition, the environmental modeling accuracy is not high, and there is a lack of comprehensive understanding of complex three-dimensional environments, which makes it easy for robots to get lost or collide in unknown or changing environments. Finally, most of the existing navigation algorithms are static planning, lack of adaptive capabilities, and difficult to cope with dynamically changing environments and task requirements. Summary of the invention

[0027] In view of the problems existing in the prior art, the present invention provides a navigation method for an intelligent navigation robot based on multi-sensor fusion.

[0028] The present invention is implemented in this way: a navigation method of an intelligent navigation robot based on multi-sensor fusion, comprising the following steps:

[0029] S1, obtaining environmental data: using lidar, visual sensors and ultrasonic sensors to simultaneously collect environmental information;

[0030] S2, data preprocessing: noise reduction, calibration and synchronization of the acquired sensor data;

[0031] S3, environmental modeling: building an environmental model based on fused sensor data;

[0032] S4, path planning: Generate the optimal navigation path using the environment model and path planning algorithm;

[0033] S5, real-time correction: Real-time position correction and dynamic obstacle avoidance are performed through sensor feedback to ensure that the robot navigates along the optimal path.

[0034] Furthermore, the data preprocessing step includes:

[0035] Filter the lidar data to remove noise points;

[0036] Perform image enhancement and feature extraction on visual sensor data;

[0037] Perform distance correction on ultrasonic sensor data to eliminate measurement errors.

[0038] Further, the environment modeling step includes:

[0039] Use lidar data to build a 2D map of the environment;

[0040] Obstacle identification and labeling through image data obtained by visual sensors;

[0041] The three-dimensional environment model is constructed by integrating the two-dimensional map and obstacle information.

[0042] Further, the path planning step includes:

[0043] The dung beetle optimization algorithm is used for global path planning to generate a preliminary navigation path;

[0044] The Dijkstra algorithm is used to locally optimize the preliminary path to ensure the smoothness and safety of the path.

[0045] Further, the real-time correction step includes:

[0046] The current position of the robot is estimated in real time through the particle filter algorithm;

[0047] Adjust navigation paths based on real-time position and sensor feedback;

[0048] An obstacle avoidance algorithm is used to replan the path when there are dynamic obstacles on the path.

[0049] Another object of the present invention is to provide an intelligent navigation robot based on multi-sensor fusion for realizing the navigation method of the intelligent navigation robot based on multi-sensor fusion, comprising:

[0050] A sensor suite, including lidar, vision sensors, and ultrasonic sensors, for simultaneously collecting environmental information;

[0051] A central processing unit for receiving data from the sensor suite and for performing environmental modeling, path planning, and real-time correction tasks;

[0052] Motion control system, used to drive the movement and steering of the robot according to the control instructions of the central processor;

[0053] A power supply system, used to provide a stable power supply to the robot;

[0054] Communication module, used for communication between the robot and the remote control system or other robots.

[0055] Furthermore, the central processing unit further comprises:

[0056] A data preprocessing module for noise reduction, calibration, and synchronization of data from the sensor suite;

[0057] Environmental modeling module, which builds environmental models based on fused sensor data;

[0058] The path planning module uses the environment model and path planning algorithm to generate the optimal navigation path;

[0059] The real-time correction module performs real-time position correction and dynamic obstacle avoidance through sensor feedback to ensure that the robot navigates along the optimal path.

[0060] Another object of the present invention is to provide a robot path planning system based on a dung beetle optimization algorithm, the system comprising:

[0061] The initial position setting module is used to initialize the initial position of the individuals of the dung beetle population by improving the Tent chaotic mapping;

[0062] The behavior mode module sets two behavior modes, obstacle-free and obstacle-involved, according to the robot's motion environment in a two-dimensional static space, and uses a specific formula to update the position of the dung beetle;

[0063] Adaptive perturbation module, used to introduce adaptive t-distribution perturbation in the process of dung beetle position update to enhance global search capability;

[0064] The breeding behavior module is used to calculate the egg-laying area and location of dung beetles during reproduction;

[0065] The foraging behavior module is used to determine the best foraging area and foraging location for dung beetles when foraging;

[0066] A stealing behavior module, used to update the position of the dung beetle according to the stealing behavior of the dung beetle;

[0067] The Cauchy mutation and reverse learning module is used to perturb and mutate the global optimal solution through Cauchy mutation and reverse learning strategies in the later stage of the algorithm;

[0068] The path planning module constructs a grid map and performs path planning based on the robot's movement in a two-dimensional static space, using the path length as the design criterion of the fitness function.

[0069] Furthermore, the system includes:

[0070] The obstacle detection module is used to define the obstacle-free environment as a free grid (white grid) and the obstacle-containing environment as a prohibited grid (black grid) when constructing a grid map;

[0071] The fitness function calculation module is used to determine the starting point and end point of the robot's planned path, initialize the middle point of the path at the robot's current row, and determine the fitness function value based on whether there are obstacles around the middle point.

[0072] Furthermore, the fitness function calculation module further comprises the following steps:

[0073] Step 1, initialize a path midpoint at the current robot row;

[0074] Step 2: Determine whether the new path point will cause the robot to encounter an obstacle;

[0075] If the new path point is an obstacle, the surrounding non-obstacle grids are selected as the new intermediate path point;

[0076] If the new path point is not an obstacle and the point is not in the existing path, add it to the path;

[0077] Step 3, repeat steps 1 and 2 until the path connects the starting point and the end point, and calculate the total length of the path as the fitness function value.

[0078] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0079] First, the existing robot navigation technology still has many shortcomings in its performance in complex environments. First, the reliability of single sensor data is low and is easily disturbed by environmental noise and obstacles, resulting in reduced navigation accuracy. Second, the existing path planning algorithm has a slow response speed in a dynamic environment and cannot avoid obstacles in real time, which affects the safety of the robot and the efficiency of task completion. In addition, the environmental modeling accuracy is not high, and there is a lack of comprehensive understanding of complex three-dimensional environments, which makes it easy for robots to get lost or collide in unknown or changing environments. Finally, most of the existing navigation algorithms are static planning, lack of adaptive capabilities, and difficult to cope with dynamically changing environments and task requirements.

[0080] The present invention significantly improves the accuracy and reliability of environmental perception by introducing multi-sensor fusion technology. It uses laser radar, visual sensors and ultrasonic sensors to comprehensively obtain environmental information, and performs noise reduction, calibration and synchronization processing on the data, overcoming the limitations of a single sensor and ensuring the accuracy and real-time nature of environmental data. Through multi-sensor fusion technology, the robot can build an accurate three-dimensional environmental model in a complex environment, effectively identify obstacles and path information, thereby improving the accuracy and safety of navigation.

[0081] The present invention uses advanced path planning algorithms, such as the dung beetle optimization algorithm and the Dijkstra algorithm, to achieve global path planning and local path optimization. Combined with the particle filter algorithm for real-time position correction, the robot can maintain high-precision navigation capabilities in a dynamic environment. When encountering dynamic obstacles, the robot can respond quickly and use the obstacle avoidance algorithm to replan the path to ensure safe obstacle avoidance and task continuity. This adaptive dynamic obstacle avoidance capability significantly improves the robot's safety and task completion efficiency, especially in complex and changing environments.

[0082] By introducing multi-sensor fusion and advanced environmental modeling technology, the present invention realizes comprehensive perception and accurate modeling of complex three-dimensional environments. The robot can dynamically update the environmental model based on the real-time collected environmental data, maintaining sensitivity and adaptability to environmental changes. At the same time, the use of adaptive navigation algorithms enables the robot to autonomously adjust navigation strategies and path planning according to task requirements and environmental changes. This highly adaptive capability enables the robot to flexibly respond to various challenges in complex and dynamic environments, significantly improving the reliability and intelligence level of the robot in practical applications.

[0083] Second, the intelligent robot navigation method provided by the present invention significantly improves and enhances the traditional Scarabaeus Optimization Algorithm (SOA) by introducing improved Tent chaos mapping, adaptive t distribution update, Cauchy mutation and reverse learning strategy. The following are the significant technical advances brought about by these technical improvements:

[0084] Tent chaos map initialization: By using Tent chaos map to initialize the population individuals, this method can not only generate a more diverse initial population, but also increase the algorithm's exploration ability, thereby finding the global optimal solution in a larger search space.

[0085] Adaptive t-distribution position update: Adaptive t-distribution is used to update the position of the rolling dung beetle. Compared with the traditional method, this update mechanism is more flexible and can adaptively adjust the search strategy according to the dynamic changes in the search process, effectively improving the search efficiency and optimization performance.

[0086] Cauchy mutation and reverse learning strategy: By introducing Cauchy mutation and reverse learning strategy, the global optimal solution is perturbed and mutated, which can not only increase the diversity of the population, but also help the algorithm jump out of the local optimum, enhance the global search ability, and increase the probability of finding a better solution.

[0087] The intelligent robot navigation method provided by the present invention significantly improves the algorithm's search capability, convergence speed and solution quality by combining multiple efficient optimization strategies, and is particularly suitable for solving complex optimization problems. These technological advances not only enhance the practicality of the algorithm, but also provide new ideas and methods for subsequent optimization algorithm research.

[0088] Third, the technical effects and technical problems solved by the intelligent robot navigation method provided by the present invention in industrial applications can be explained from the following aspects:

[0089] Technical effect:

[0090] 1) Improve optimization efficiency: By initializing the population using the improved Tent chaotic map, the convergence speed of the algorithm is accelerated, so that the solution to the problem can be found more quickly under the same computing resource conditions.

[0091] 2) Enhanced global search capability: The introduction of adaptive t-distribution update and Cauchy mutation strategy enables the algorithm to effectively explore a wider search space, increasing the probability of finding the global optimal solution, especially for high-dimensional complex optimization problems.

[0092] 3) Improve the quality of the solution: The application of the reverse learning strategy enables the algorithm to self-learn and adjust during the iteration process, effectively avoiding falling into the local optimal solution and thus obtaining a better solution.

[0093] 4) Increase the robustness of the algorithm: The fusion of multiple strategies not only improves the search efficiency and optimization performance of the algorithm, but also makes the algorithm more robust, that is, it can maintain good performance on different types of optimization problems.

[0094] Technical issues solved:

[0095] 1) Premature convergence problem: In traditional optimization methods, the algorithm is prone to fall into the local optimal solution and cannot jump out, resulting in premature convergence. The intelligent robot navigation method effectively solves this problem by introducing new mutation and learning strategies.

[0096] 2) Insufficient exploration of search space: For some complex problems with large search spaces, traditional algorithms often fail to find satisfactory solutions due to limited search capabilities. This method expands the search range and improves the comprehensiveness of the search through chaotic mapping and adaptive update mechanism.

[0097] 3) Low optimization efficiency: When dealing with high-dimensional or computationally intensive industrial optimization problems, traditional optimization algorithms often suffer from low computational efficiency. This optimization method significantly improves computational efficiency by improving the initialization and update mechanisms.

[0098] 4) Insufficient algorithm robustness: For optimization problems with different characteristics, traditional algorithms need to adjust parameters in a targeted manner to achieve better results. Intelligent robot navigation methods enhance the adaptability and robustness of algorithms by introducing multiple strategies, reducing dependence on external parameter adjustments.

[0099] The intelligent robot navigation method provided by the present invention provides an efficient and robust solution for solving complex industrial optimization problems, and is particularly suitable for application scenarios that have high requirements on optimization speed and solution quality, such as industrial design optimization, production scheduling, resource allocation and other fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] Figure 1 is a flow chart of a navigation method of an intelligent navigation robot based on multi-sensor fusion provided by an embodiment of the present invention;

[0101] Figure 2 is a schematic diagram of the structure of an intelligent navigation robot based on multi-sensor fusion provided by an embodiment of the present invention;

[0102] Figure 3 is a structural diagram of a robot path planning system based on a dung beetle optimization algorithm provided in an embodiment of the present invention;

[0103] Figure 4 It is the optimal fitness value diagram provided by the embodiment of the present invention.

[0104] Figure 5 This is a path planning effect diagram provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0105] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0106] like Figure 1 As shown, the intelligent navigation robot navigation method based on multi-sensor fusion provided by an embodiment of the present invention includes the following steps:

[0107] S1, obtaining environmental data: using lidar, visual sensors and ultrasonic sensors to simultaneously collect environmental information;

[0108] S2, data preprocessing: noise reduction, calibration and synchronization of the acquired sensor data;

[0109] S3, environmental modeling: building an environmental model based on fused sensor data;

[0110] S4, path planning: Generate the optimal navigation path using the environment model and path planning algorithm;

[0111] S5, real-time correction: Real-time position correction and dynamic obstacle avoidance are performed through sensor feedback to ensure that the robot navigates along the optimal path.

[0112] The data preprocessing step includes:

[0113] Filter the lidar data to remove noise points;

[0114] Perform image enhancement and feature extraction on visual sensor data;

[0115] Perform distance correction on ultrasonic sensor data to eliminate measurement errors.

[0116] The environmental modeling step includes:

[0117] Use lidar data to build a 2D map of the environment;

[0118] Obstacle identification and labeling through image data obtained by visual sensors;

[0119] The three-dimensional environment model is constructed by integrating the two-dimensional map and obstacle information.

[0120] The path planning step includes:

[0121] The dung beetle optimization algorithm is used for global path planning to generate a preliminary navigation path;

[0122] The Dijkstra algorithm is used to locally optimize the preliminary path to ensure the smoothness and safety of the path.

[0123] The real-time correction step comprises:

[0124] The current position of the robot is estimated in real time through the particle filter algorithm;

[0125] Adjust navigation paths based on real-time position and sensor feedback;

[0126] An obstacle avoidance algorithm is used to replan the path when there are dynamic obstacles on the path.

[0127] like Figure 2 As shown, the intelligent navigation robot based on multi-sensor fusion provided by the embodiment of the present invention includes:

[0128] A sensor suite, including lidar, vision sensors, and ultrasonic sensors, for simultaneously collecting environmental information;

[0129] A central processing unit for receiving data from the sensor suite and for performing environmental modeling, path planning, and real-time correction tasks;

[0130] Motion control system, used to drive the movement and steering of the robot according to the control instructions of the central processor;

[0131] A power supply system, used to provide a stable power supply to the robot;

[0132] Communication module, used for communication between the robot and the remote control system or other robots.

[0133] The central processing unit further comprises:

[0134] A data preprocessing module for noise reduction, calibration, and synchronization of data from the sensor suite;

[0135] Environmental modeling module, which builds environmental models based on fused sensor data;

[0136] The path planning module uses the environment model and path planning algorithm to generate the optimal navigation path;

[0137] The real-time correction module performs real-time position correction and dynamic obstacle avoidance through sensor feedback to ensure that the robot navigates along the optimal path.

[0138] like Figure 3 As shown, a robot path planning system based on a dung beetle optimization algorithm provided in an embodiment of the present invention comprises:

[0139] The initial position setting module is used to initialize the initial position of the individuals of the dung beetle population by improving the Tent chaotic mapping;

[0140] The behavior mode module sets two behavior modes, obstacle-free and obstacle-involved, according to the robot's motion environment in a two-dimensional static space, and uses a specific formula to update the position of the dung beetle;

[0141] Adaptive perturbation module, used to introduce adaptive t-distribution perturbation in the process of dung beetle position update to enhance global search capability;

[0142] The breeding behavior module is used to calculate the egg-laying area and location of dung beetles during reproduction;

[0143] The foraging behavior module is used to determine the best foraging area and foraging location for dung beetles when foraging;

[0144] A stealing behavior module, used to update the position of the dung beetle according to the stealing behavior of the dung beetle;

[0145] The Cauchy mutation and reverse learning module is used to perturb and mutate the global optimal solution through Cauchy mutation and reverse learning strategies in the later stage of the algorithm;

[0146] The path planning module constructs a grid map and performs path planning based on the robot's movement in a two-dimensional static space, using the path length as the design criterion of the fitness function.

[0147] Furthermore, the system includes:

[0148] The obstacle detection module is used to define the obstacle-free environment as a free grid (white grid) and the obstacle-containing environment as a prohibited grid (black grid) when constructing a grid map;

[0149] The fitness function calculation module is used to determine the starting point and end point of the robot's planned path, initialize the middle point of the path at the robot's current row, and determine the fitness function value based on whether there are obstacles around the middle point.

[0150] The fitness function calculation module also includes the following steps:

[0151] Step 1, initialize a path midpoint at the current robot row;

[0152] Step 2: Determine whether the new path point will cause the robot to encounter an obstacle;

[0153] If the new path point is an obstacle, the surrounding non-obstacle grids are selected as the new intermediate path point;

[0154] If the new path point is not an obstacle and the point is not in the existing path, add it to the path;

[0155] Step 3, repeat steps 1 and 2 until the path connects the starting point and the end point, and calculate the total length of the path as the fitness function value.

[0156] Application Example 1: Warehouse Management Robot

[0157] In large warehouses, the storage and retrieval of items requires a lot of manual operations, which is inefficient and error-prone. In order to improve the efficiency of warehouse management, an intelligent robot navigation method based on multi-sensor fusion is introduced.

[0158] 1) Environmental data acquisition:

[0159] Warehouse robots are equipped with lidar, visual sensors, and ultrasonic sensors, which simultaneously collect environmental data within the warehouse, including shelf locations, obstacles, and path information.

[0160] 2) Data preprocessing:

[0161] The lidar data is filtered to remove noise points and ensure data accuracy.

[0162] Vision sensors enhance and extract features from image data to identify cargo labels and locations.

[0163] Ultrasonic sensor data is used for distance correction to eliminate measurement errors and ensure measurement accuracy.

[0164] 3) Environmental Modeling:

[0165] Use LiDAR data to build a 2D environment map of the warehouse, marking the locations of shelves and aisles.

[0166] Obstacles are identified and annotated through image data acquired by visual sensors to generate a comprehensive three-dimensional environment model.

[0167] 4) Path planning:

[0168] The dung beetle optimization algorithm is used for global path planning to generate the robot's preliminary navigation path from the warehouse entrance to the target shelf.

[0169] The Dijkstra algorithm is used to locally optimize the preliminary path to ensure the smoothness and safety of the path and avoid collisions.

[0170] 5) Real-time correction and dynamic obstacle avoidance:

[0171] The robot estimates its current position in real time through a particle filter algorithm and makes path corrections based on sensor feedback information.

[0172] When encountering dynamic obstacles on the path (such as other moving robots or workers), an obstacle avoidance algorithm is used to replan the path to ensure safe arrival at the target location.

[0173] Through this method, warehouse robots can store and retrieve goods efficiently and accurately, greatly improving the efficiency of warehouse management and reducing the error rate of manual operations.

[0174] Application Example 2: Home Service Robot

[0175] In order to improve the convenience and safety of family life, an intelligent robot navigation method based on multi-sensor fusion is introduced for home service robots.

[0176] 1) Environmental data acquisition:

[0177] The home service robot is equipped with lidar, visual sensors and ultrasonic sensors that collect home environment data, including room layout, furniture location and obstacle information.

[0178] 2) Data preprocessing:

[0179] The lidar data is filtered to remove noise points and ensure the accuracy of the data.

[0180] Vision sensors enhance and extract features from image data to identify the location and status of family members.

[0181] Ultrasonic sensor data is used for distance correction to eliminate measurement errors and ensure measurement accuracy.

[0182] 3) Environmental Modeling:

[0183] Use lidar data to build a 2D map of your home, annotating the locations of rooms and furniture.

[0184] Obstacles are identified and annotated through image data acquired by visual sensors to generate a comprehensive three-dimensional environment model.

[0185] 4) Path planning:

[0186] The dung beetle optimization algorithm is used for global path planning to generate the robot's preliminary navigation path within the home, such as from the living room to the kitchen.

[0187] The Dijkstra algorithm is used to locally optimize the preliminary path to ensure the smoothness and safety of the path and avoid collisions with furniture or other obstacles.

[0188] 5) Real-time correction and dynamic obstacle avoidance:

[0189] The robot estimates its current position in real time through a particle filter algorithm and makes path corrections based on sensor feedback information.

[0190] When encountering dynamic obstacles on the path (such as pets or family members), obstacle avoidance algorithms are used to re-plan the path to ensure safe navigation.

[0191] Through this method, the home service robot can move efficiently and safely in a complex home environment to perform tasks such as cleaning and delivering items, significantly improving the convenience and safety of family life.

[0192] The working principle of the method provided by the embodiment of the present invention.

[0193] Step 1: Get environment data

[0194] In the intelligent robot navigation method, the first step is to obtain environmental data. The robot is equipped with lidar, visual sensors and ultrasonic sensors, which work simultaneously to collect information about the surrounding environment:

[0195] LiDAR: It generates 2D or 3D point cloud data of the environment by emitting and receiving laser beams to measure the distance to surrounding objects.

[0196] Vision sensor: A camera or depth camera acquires images of the environment, performs image processing, and identifies obstacles, paths, and other features.

[0197] Ultrasonic sensor: It measures distance and provides close-range obstacle detection information by emitting and receiving ultrasonic signals.

[0198] Step 2: Data Preprocessing

[0199] After obtaining the sensor data, it is necessary to preprocess the data to ensure its accuracy and reliability:

[0200] Filtering: Filter the lidar data to remove noise points and retain effective environmental information.

[0201] Image enhancement and feature extraction: Perform image enhancement on visual sensor data to improve image quality, and perform feature extraction to identify useful environmental features such as obstacles and paths.

[0202] Distance correction: Perform distance correction on ultrasonic sensor data to eliminate measurement errors and ensure the accuracy of measurement results.

[0203] Step 3: Environment Modeling

[0204] Based on the preprocessed sensor data, the robot builds a model of the environment:

[0205] 2D environment map: Generate a 2D environment map using lidar data to identify obstacles and paths on the ground.

[0206] Obstacle identification and labeling: Obstacle identification is performed through visual sensor data, and the location and type of obstacles are marked.

[0207] Three-dimensional environment model: Combine the two-dimensional map and obstacle information to build a three-dimensional environment model to more comprehensively reflect the environment structure.

[0208] Step 4: Path Planning

[0209] Based on the environmental model, the robot performs path planning and generates the optimal navigation path:

[0210] Global path planning: The dung beetle optimization algorithm is used for global path planning to determine the initial navigation path from the starting point to the target point. The dung beetle optimization algorithm uses heuristic functions to quickly find the optimal path.

[0211] Local path optimization: Use the Dijkstra algorithm to locally optimize the preliminary path to ensure the smoothness and safety of the path and avoid collisions and unnecessary detours.

[0212] Step 5: Real-time correction

[0213] During navigation, the robot uses sensor feedback information to perform real-time position correction and dynamic obstacle avoidance:

[0214] Position estimation: The current position of the robot is estimated in real time through the particle filter algorithm, and the position information is updated in combination with the sensor data.

[0215] Path adjustment: Based on real-time position and sensor feedback information, the navigation path is adjusted to ensure that the robot moves along the optimal path.

[0216] Dynamic obstacle avoidance: Using an obstacle avoidance algorithm, when there are dynamic obstacles on the path, the path is replanned to ensure that the robot can safely avoid obstacles and continue to navigate to the target location.

[0217] Through the above steps, the robot can navigate efficiently and safely in complex and dynamic environments. Multi-sensor fusion technology improves the accuracy and reliability of environmental perception, and path planning and real-time correction algorithms ensure the robot's adaptability and safety in dynamic environments. In summary, the intelligent robot navigation method of the present invention significantly improves the robot's navigation performance and task completion efficiency in practical applications.

[0218] The detailed working principle of the system provided by the embodiment of the present invention.

[0219] Initial position setting module: In the robot path planning system, the first step is to initialize the initial position of the dung beetle population. The improved Tent chaotic mapping algorithm is used to generate the initial population position with chaotic characteristics to ensure that the population is evenly distributed in the search space and enhance the global search capability. The adjustment factor and chaotic factor setting of the Tent chaotic mapping make the initial position of the population have good traversability and diversity.

[0220] Behavior mode module: According to the robot's motion environment, two behavior modes are set: barrier-free and barrier-free:

[0221] 1. Accessible behavior mode:

[0222] The robot moves in an obstacle-free environment and uses a specific formula to update the position of the dung beetle. The formula contains the deflection coefficient, constant value and global worst position, which is used to simulate the movement trajectory of the dung beetle and ensure its efficient movement in an obstacle-free environment.

[0223] 2. Impaired behavior patterns:

[0224] The robot moves in an environment with obstacles, uses a specific formula to update the position of the dung beetle, and limits the rolling direction to a certain range. If the rolling direction does not meet the conditions, the position will not be updated to ensure that the dung beetle can effectively avoid obstacles.

[0225] Adaptive perturbation module:

[0226] In order to improve the global search capability, an adaptive t-distribution perturbation is introduced in the process of the dung beetle's position update. The adaptive t-distribution has a heavy-tailed characteristic and can perform large-step mutations in the search space, effectively preventing the dung beetle from falling into the local optimal solution and enhancing the global search capability.

[0227] Reproduction behavior module: In the reproduction behavior of dung beetles, the egg-laying area and egg-laying position are calculated: the current local optimal position represents the optimal solution for the dung beetle at the current position. and represent the upper and lower bounds of the reproduction area, respectively. and represent the upper and lower bounds of the optimization problem, respectively, to ensure that the egg-laying position is within a reasonable range. The reproduction behavior module dynamically adjusts the egg-laying area to ensure population diversity and comprehensive coverage of the search space.

[0228] Foraging behavior module: Calculate the optimal foraging area and foraging position of the dung beetle in its foraging behavior: In the foraging behavior, determine the lower and upper limits of the optimal foraging area by analyzing the global optimal position. Dynamically adjust the foraging position according to the requirements of the environment and the optimization problem to ensure that the foraging behavior effectively improves the quality of the solution.

[0229] Stealing Behavior Module: Update the position of dung beetles according to their stealing behavior: During foraging, dung beetles update their own positions by stealing food sources from other dung beetles. Use a specific formula to calculate the position update to ensure that the stealing behavior plays a role in the global search.

[0230] Cauchy mutation and reverse learning module: In the later stage of the dung beetle optimization algorithm, the Cauchy mutation and reverse learning strategies are introduced to perturb and mutate the global optimal solution: Cauchy mutation strategy: Utilize the heavy-tail characteristics of the Cauchy distribution to perform large-step mutation on the global optimal solution to improve the global search capability.

[0231] Reverse learning strategy: By reversely exploring the current optimal solution, the search range of solutions is expanded and the diversity of the solution space is increased.

[0232] Path planning module: Based on the robot's movement in a two-dimensional static space, a grid map is constructed for path planning. The path length is used as the design criterion of the fitness function to optimize the path planning: a grid map is used to represent obstacles and open areas in a two-dimensional static space. Based on the dung beetle optimization algorithm, the optimization strategies provided by each module are used to find the optimal path. By calculating the path length, the fitness of the path planning is evaluated to ensure that the robot can reach the target location efficiently and safely.

[0233] Through the collaborative work of the above modules, the robot path planning system based on the dung beetle optimization algorithm can efficiently plan paths in complex and static environments. The initial position setting module ensures population diversity, the adaptive perturbation and Cauchy mutation strategies enhance global search capabilities, and the behavior pattern module and the foraging and reproduction behavior module improve search efficiency and accuracy. Finally, the path planning module generates the optimal path based on the optimization results, significantly improving the performance and reliability of the robot path planning.

[0234] The system provided by the embodiment of the present invention also includes the following modules.

[0235] Initial position setting module: In the robot path planning system, the first step is to initialize the initial position of the dung beetle population. The improved Tent chaotic mapping algorithm is used to generate the initial population position with chaotic characteristics to ensure that the population is evenly distributed in the search space and enhance the global search capability. The adjustment factor and chaotic factor setting of the Tent chaotic mapping make the initial position of the population have good traversability and diversity.

[0236] Obstacle Detection Module: When building a grid map, the obstacle detection module divides the environment into free grids and prohibited grids based on sensor data:

[0237] Free grid: An obstacle-free environment, represented by a white grid, where the robot can pass freely.

[0238] No-entry grid: An environment with obstacles, represented by a black grid, through which the robot cannot pass.

[0239] Behavior pattern module:

[0240] According to the robot's motion environment, two behavior modes are set: barrier-free and barrier-free:

[0241] 1. Accessible behavior mode:

[0242] The robot moves in an obstacle-free environment and uses a specific formula to update the position of the dung beetle. The formula contains the deflection coefficient, constant value and global worst position, which is used to simulate the movement trajectory of the dung beetle and ensure its efficient movement in an obstacle-free environment.

[0243] 2. Impaired behavior patterns:

[0244] The robot moves in an environment with obstacles, uses a specific formula to update the position of the dung beetle, and limits the rolling direction to a certain range. If the rolling direction does not meet the conditions, the position will not be updated to ensure that the dung beetle can effectively avoid obstacles.

[0245] Adaptive perturbation module:

[0246] In order to improve the global search capability, an adaptive t-distribution perturbation is introduced in the process of the dung beetle's position update. The adaptive t-distribution has a heavy-tailed characteristic and can perform large-step mutations in the search space, effectively preventing the dung beetle from falling into the local optimal solution and enhancing the global search capability.

[0247] Reproduction Behavior Module:

[0248] In the reproductive behavior of dung beetles, calculate their egg-laying area and egg-laying location:

[0249] The current local optimal position represents the optimal solution for the dung beetle at its current position.

[0250] and represent the upper and lower bounds of the breeding area, respectively.

[0251] and represent the upper and lower bounds of the optimization problem, respectively, ensuring that the spawning position is within a reasonable range.

[0252] The reproductive behavior module ensures population diversity and comprehensive coverage of the search space by dynamically adjusting the spawning area.

[0253] Foraging behavior module: Calculate the optimal foraging area and foraging position of the dung beetle during its foraging behavior: During foraging behavior, determine the lower and upper limits of the optimal foraging area by analyzing the global optimal position.

[0254] The foraging positions are adjusted dynamically according to the requirements of the environment and the optimization problem, ensuring that the foraging behavior effectively improves the quality of the solution.

[0255] Stealing behavior module: Update the position of dung beetles according to their stealing behavior: During the foraging process, dung beetles update their own positions by stealing the food sources of other dung beetles.

[0256] The position update is calculated using a specific formula to ensure that stealing behavior works in the global search.

[0257] Cauchy mutation and reverse learning module: In the later stage of the dung beetle optimization algorithm, Cauchy mutation and reverse learning strategies are introduced to perturb and mutate the global optimal solution:

[0258] Cauchy mutation strategy: Utilize the heavy-tailed characteristics of the Cauchy distribution to perform large-step mutations on the global optimal solution to improve the global search capability.

[0259] Reverse learning strategy: By reversely exploring the current optimal solution, the search range of solutions is expanded and the diversity of the solution space is increased.

[0260] Path planning module:

[0261] According to the movement of the robot in a two-dimensional static space, a grid map is constructed for path planning:

[0262] Grid map: White grids represent free areas and black grids represent obstacle areas.

[0263] Path length as fitness function: Path length is used as the design criterion of the fitness function to optimize path planning.

[0264] Fitness function calculation module:

[0265] The fitness function calculation module is used to evaluate the quality of the planned path:

[0266] 1. Determine the starting point and end point: Determine the starting point and end point of the robot's planned path.

[0267] 2. Path midpoint initialization: Initialize a path midpoint at the current line of the robot.

[0268] 3. Path obstacle detection: Determine whether there are obstacles on the path.

[0269] If there are obstacles: assign a larger value to the fitness function to penalize infeasible paths.

[0270] If there are no obstacles: perform a midpoint neighborhood search to find the best path.

[0271] If the new grid is an obstacle, the surrounding non-obstacle grids are taken as intermediate path points.

[0272] If it is not an obstacle and the point is not in the path, add it to the path.

[0273] Through the collaborative work of the above modules, the robot path planning system based on the dung beetle optimization algorithm can efficiently plan paths in complex and static environments. The initial position setting module ensures population diversity, the adaptive perturbation and Cauchy mutation strategies enhance global search capabilities, and the behavior pattern module and the foraging and reproduction behavior module improve search efficiency and accuracy. The obstacle detection module ensures that path planning takes into account obstacles in the environment, and the fitness function calculation module optimizes the path quality through path evaluation. Finally, the path planning module generates the optimal path based on the optimization results, which significantly improves the performance and reliability of the robot path planning.

[0274] As shown, an intelligent robot navigation method provided by an embodiment of the present invention comprises the following steps:

[0275] Initialization step: using Tent chaotic mapping to initialize the initial positions of the dung beetle population individuals, the Tent chaotic mapping includes a regulating factor and a chaotic factor, and the initial population position with chaotic characteristics is generated by setting the values ​​of the two factors;

[0276] Behavior pattern updating step: the behavior pattern of the ball beetle is divided into an obstacle-free behavior pattern and an obstacle behavior pattern, the position of the ball beetle is updated according to the current environmental conditions and the corresponding behavior pattern, and an adaptive t distribution is introduced in the updating process for perturbation;

[0277] In the barrier-free behavior mode, the position update formula of the rolling ball dung beetle is used to update the position of the i-th dung beetle at the t-th iteration, where the update formula includes the deflection coefficient, constant value, natural coefficient and global worst position;

[0278] In the obstacle behavior mode, the position of the rolling ball dung beetle is updated using a specific position update formula, and the rolling direction is restricted to a certain range. If the rolling direction does not meet the conditions, the position update will not be performed.

[0279] Global optimal solution perturbation step: Introduce the Cauchy mutation strategy and reverse learning strategy to perturb and mutate the global optimal solution. The Cauchy mutation strategy uses the heavy-tailed characteristics of the Cauchy distribution to perform large-step mutations in the search space, and the reverse learning strategy expands the search range of the solution and increases the diversity of the solution space by reversely exploring the current optimal solution.

[0280] Performance evaluation and selection step: In each iteration, the performance of all candidate solutions generated by the Cauchy mutation and reverse learning strategies is evaluated, and a new global optimal solution or updated solution set is selected based on the evaluation results;

[0281] Iteration termination step: Continue the above steps until the termination condition of the algorithm is met, such as reaching the preset maximum number of iterations or the degree of improvement of the solution is lower than the preset threshold, thereby outputting the optimization result.

[0282] In S101 provided in the embodiment of the present invention, the initial positions of the individuals in the dung beetle population are initialized using the Tent chaotic mapping, and the formula of the Tent chaotic mapping is as follows:

[0283]

[0284] In the formula: ζ∈(0,1] is the adjustment factor, κ∈(0,2] is the chaos factor, ζ is 1, κ is 0.5;

[0285] The initial position of the population after mapping is:

[0286] z i =lb+(ub-lb)×pop i ;

[0287] In the formula: ub and lb are the upper and lower limits of each individual respectively;

[0288] In S102 provided in the embodiment of the present invention, in the barrier-free behavior mode, the position update formula of the rolling dung beetle is as follows:

[0289]

[0290] Where t represents the current iteration number, represents the position of the i-th dung beetle at the t-th iteration, k is the deflection coefficient, k∈(0,0.2], b represents a constant value belonging to (0,1), α is a natural coefficient with a value of 1 or 1, where 1 represents a deviation from the original direction; Z w represents the global worst position, and Δz is used to simulate the intensity variation of light.

[0291] In the obstacle behavior mode of the embodiment of the present invention, the position update formula of the rolling dung beetle is as follows:

[0292]

[0293] The scroll direction θ is only considered to be between [0, π]. If θ is equal to 0, π / 2 or π, no position update is performed;

[0294] An adaptive t distribution is introduced to perturb the position update of the rolling dung beetle. The probability density formula of the adaptive t distribution is as follows:

[0295]

[0296] The boundary of the egg-laying area of ​​the dung beetle provided in the embodiment of the present invention changes dynamically when it reproduces, and the formula is as follows:

[0297] Lb * =max(Z * ×(1-R),Lb);

[0298] Ub * =min(Z * ×(1+R),Ub);

[0299] Among them, Z * represents the current local optimal position, Lb * and Ub * Respectively represent the upper and lower limits of the breeding area, R = 1-t / T max , where T max represents the maximum number of iterations, Lb and Ub represent the upper and lower bounds of the optimization problem respectively; therefore, the spawning position will change dynamically, and the formula is as follows:

[0300]

[0301] in, is the egg-laying position of the ith female dung beetle in the tth iteration, b1 and b 2 represents two independent random vectors of size 1×D, where D represents the dimension of the optimization problem;

[0302] Similarly, the optimal foraging area in foraging behavior also changes dynamically, and the formula is as follows:

[0303] Lb b =max(Z b ×(1-R),Lb);

[0304] Ub b =min(Z b ×(1+R),Ub);

[0305] Among them, Z b represents the global optimal position, Lb b and Ub b Respectively represent the lower and upper limits of the optimal foraging area; the position update formula of the foraging dung beetle is:

[0306]

[0307] Among them, C 1 represents a random number that follows a normal distribution, and C 2 represents a random vector belonging to (0,1);

[0308] The formula for updating the position of the stealing dung beetle is:

[0309]

[0310] Among them, Z b is the best food source; g is a random vector of size 1×D that obeys a normal distribution, and S represents a constant value.

[0311] The mathematical definition of the Cauchy mutation strategy provided in the embodiment of the present invention is as follows:

[0312]

[0313] Where: Z best For elite individual positions, is the position of the elite individual after mutation, σ is the standard deviation of the Cauchy mutation strategy; Cauchy(0,σ 2 ) is a random variable that follows a Cauchy distribution.

[0314] A robot path planning system provided by an embodiment of the present invention includes:

[0315] An initialization module, which uses Tent chaotic mapping to initialize the initial positions of the individuals in the dung beetle population, wherein the formula of the Tent chaotic mapping includes a regulating factor and a chaotic factor to generate an initial population position with chaotic characteristics;

[0316] The behavior pattern updating module divides the behavior pattern of the rolling dung beetle into an obstacle-free behavior pattern and an obstacle-related behavior pattern, updates the position of the dung beetle according to the current environment and behavior pattern, and introduces an adaptive t-distribution for perturbation during the updating process;

[0317] The global optimal solution perturbation module introduces the Cauchy mutation strategy and the reverse learning strategy to perturb and mutate the global optimal solution. The Cauchy mutation strategy uses the heavy-tailed characteristics of the Cauchy distribution to perform large-step mutations, and the reverse learning strategy expands the search range and increases the diversity of the solution space by reversely exploring the current optimal solution.

[0318] Performance evaluation and selection module, which evaluates the performance of all candidate solutions generated by Cauchy mutation and reverse learning strategies in each iteration, and selects a new global optimal solution or updates the solution set based on the evaluation results;

[0319] Iteration control module, which is used to control the number of iterations of the optimization process and stop the iteration and output the optimization result when the preset termination condition is met, such as reaching the maximum number of iterations or the improvement degree of the solution is lower than the preset threshold;

[0320] The user interface allows the user to input the parameters of the optimization problem, display the real-time status of the optimization process, and output the final optimization results.

[0321] Implementation Plan 1: Economic Dispatch Problem of Power System

[0322] Background: Economic Load Dispatch (ELD) of power system refers to adjusting the output power of generator sets to minimize the total power generation cost of the system while meeting the system load demand and operating constraints.

[0323] 1) Application of mapping module: The improved Tent chaotic mapping is used to initialize the initial positions of the dung beetle population individuals. This method increases the diversity of the population through chaotic variables and avoids the algorithm falling into local optimality in the early stage.

[0324] 2) Application of update module: The position of the rolling dung beetle is updated through adaptive t-distribution. This strategy takes into account the interaction between individuals and environmental feedback, and enhances the global exploration and local development capabilities of the search process.

[0325] 3) Introduce module application: Introduce Cauchy mutation and reverse learning strategies to perturb and mutate the global optimal solution, so as to jump out of the local optimum and enhance the global search ability of the algorithm.

[0326] 4) Optimized execution: The intelligent robot navigation method is executed on a computer device, and the processor runs an algorithm according to a stored computer program to solve the ELD problem and find the generator set output power configuration with the lowest cost.

[0327] Implementation 2: Machine Learning Model Parameter Optimization

[0328] Background: In the field of machine learning, optimizing model parameters is a key step to improve model performance, which is usually a high-dimensional and non-convex optimization problem.

[0329] 1) Mapping module application: Tent chaotic mapping is used to initialize the dung beetle population and provide an initial solution set for parameter optimization of the machine learning model. This initialization method is conducive to exploring different areas of the parameter space.

[0330] 2) Application of update module: The position of the dung beetle is updated using adaptive t-distribution. This process simulates the search behavior of the dung beetle when rolling the ball. The search step size is adaptively adjusted to balance the exploration and development capabilities.

[0331] 3) Introducing module applications: Increase the diversity of the search process through Cauchy mutation and reverse learning strategies. These strategies help the algorithm escape from the local optimum and better approach the global optimal solution.

[0332] 4) Optimization execution: The optimization system is deployed on the information data processing terminal, and the intelligent robot navigation method is executed to automatically adjust the parameters of the machine learning model to obtain the optimal model performance.

[0333] These two implementation schemes demonstrate the application potential of the robot path planning system in different fields. Through the close collaboration of mapping, updating and introducing modules, complex problems such as economic scheduling and model parameter optimization are effectively solved, and the optimization efficiency and accuracy are improved. The specific implementation method of the present invention is as follows:

[0334] The idea of ​​the dung beetle optimization method is mainly based on the four behaviors of dung beetles: rolling, reproduction, foraging and stealing.

[0335] First, the tent chaotic map is used to initialize the initial position of the dung beetle population individuals. The formula is as follows:

[0336]

[0337] Wherein: ζ∈(0,1] is the adjustment factor, κ∈(0,2] is the chaos factor, and in the present invention, ζ is 1 and κ is 0.5.

[0338] The initial position of the population after mapping is

[0339] z i =lb+(ub-lb)×pop i

[0340] In the formula: ub and lb are the upper and lower limits of each individual.

[0341] The behavior patterns of the ball-rolling dung beetle are divided into two modes: barrier-free and barrier-free.

[0342] In barrier-free behavior mode, the position update formula of the rolling dung beetle is as follows:

[0343]

[0344] Where t represents the current iteration number, represents the position of the i-th dung beetle at the t-th iteration, k is the deflection coefficient, k∈(0,0.2], b represents a constant value belonging to (0,1), and α is a natural coefficient with a value of 1 or 1, where 1 represents a deviation from the original direction. Z w represents the global worst position, and Δz is used to simulate the intensity variation of light.

[0345] In the obstacle behavior mode, the position update formula of the rolling dung beetle is as follows:

[0346]

[0347] The scroll direction θ is only considered to be between [0, π]. If θ is equal to 0, π / 2 or π, no position update is performed.

[0348] In order to improve the global exploration ability of the algorithm, an adaptive t-distribution perturbation is introduced in the position update of the rolling dung beetle to improve the global search ability.

[0349] The probability density formula of t distribution is as follows:

[0350]

[0351] The boundaries of the dung beetle's egg-laying area change dynamically when it reproduces. The formula is as follows:

[0352] Lb * =max(Z * ×(1-R),Lb)

[0353] Ub * =min(Z * ×(1+R),Ub)

[0354] Among them, Z * represents the current local optimal position, Lb * and Ub* Respectively represent the upper and lower limits of the breeding area, R = 1-t / T max , where T max represents the maximum number of iterations, Lb and Ub represent the upper and lower bounds of the optimization problem. Therefore, the spawning position will change dynamically, and the formula is as follows:

[0355]

[0356] in, is the egg-laying position of the ith female dung beetle in the tth iteration, b 1 and b 2 denote two independent random vectors of size 1×D, where D represents the dimension of the optimization problem.

[0357] Similarly, the optimal foraging area in foraging behavior also changes dynamically, and the formula is as follows:

[0358] Lb b =max(Z b ×(1-R),Lb)

[0359] Ub b =min(Z b ×(1+R),Ub)

[0360] Among them, Z b represents the global optimal position, Lb b and Ub b Represent the lower and upper limits of the optimal foraging area respectively. The position update formula of the foraging dung beetle is:

[0361]

[0362] Among them, C 1 represents a random number that follows a normal distribution, and C 2 represents a random vector belonging to (0,1).

[0363] The formula for updating the position of the stealing dung beetle is:

[0364]

[0365] Among them, Z b is the best food source. g is a 1×D random vector with a normal distribution, and S represents a constant value.

[0366] In the later stage of the dung beetle optimization method, dung beetles gradually approach the optimal individual, which will lead to the problem of lack of population diversity and premature convergence of the algorithm. To solve this problem, the present invention introduces Cauchy mutation and reverse learning strategies to perturb and mutate the global optimal solution. The algorithm selects the individual with the best current fitness for mutation, then compares the positions before and after the mutation, and selects the better position to enter the next iteration. The mathematical definition of the Cauchy mutation strategy is as follows:

[0367]

[0368] Where: Z best For elite individual positions, is the position of the elite individual after mutation, and σ is the standard deviation of the Cauchy mutation strategy. 2 ) is a random variable that follows a Cauchy distribution.

[0369] Define algorithm parameters: set the maximum number of iterations to 500 and the number of dung beetle populations to 30. To eliminate the contingency of the experiment, the mean and standard deviation of the results of the present invention are compared with those of the DBO and particle swarm algorithms, and each algorithm is run independently 100 times. Three groups of benchmark test functions are selected for simulation experiments, and the specific function information is shown in Table 1.

[0370] Table 1 Benchmark function information

[0371]

[0372] where F 1 is a single-peak test function used to evaluate the optimization performance of the algorithm; F 2 is a multi-peak test function used to evaluate the ability of the algorithm to avoid local optimality; F 3 It is a fixed-dimensional multimodal test function used to evaluate the performance of the algorithm in handling high-dimensional problems.

[0373] Table 2 shows the average values ​​and standard deviations of the three algorithms after running 100 times independently on different test functions under the benchmark test function. The smaller the average value, the better the optimization performance of the algorithm; the smaller the standard deviation, the higher the optimization stability of the algorithm.

[0374] As can be seen from Table 2, for unimodal, multimodal and fixed-dimensional multimodal test functions, MDBO outperforms the other two comparison algorithms in both optimization accuracy and search stability. 1 On the benchmark function, the mean and standard deviation of MDBO are both 0, indicating that the algorithm converges to the theoretical optimal solution, showing its strong optimization ability and robustness. 2 On the benchmark function, MDBO's performance is 5 to 6 orders of magnitude higher than other comparison algorithms, demonstrating its strong ability to avoid local optimality. 3On the benchmark function, MDBO can also obtain the optimal value, indicating that its ability to handle high-dimensional problems is better than the other two methods.

[0375] Table 2 Experimental results of MDBO and other two comparison algorithms on three groups of test functions

[0376]

[0377] In order to intuitively demonstrate the algorithm performance of MDBO, the convergence curves of the best fitness value of MDBO and the other two comparison algorithms under the benchmark test function are plotted as the number of iterations changes.

[0378] from Figure 4 It can be seen that MDBO has higher optimization accuracy, faster solution speed and stronger ability to avoid local optimality than PSO and DBO.

[0379] The multi-strategy evolutionary dung beetle optimization algorithm proposed in the present invention can be applied but not limited to:

[0380] (1) Unmanned platform motion control, path planning, and obstacle avoidance;

[0381] (2) Sensor deployment and location optimization;

[0382] (3) Hyperparameter optimization and neural network weight optimization;

[0383] (4) Task allocation and combinatorial optimization problems;

[0384] In the unmanned platform motion control, path planning, and obstacle avoidance problems, compared with traditional methods, the present invention does not need to establish complex mathematical models, performs multi-objective optimization based on the objective function, dynamically adjusts the path planning scheme according to the surrounding environment, can simultaneously and effectively handle multiple constraints, and has good real-time performance.

[0385] In sensor deployment and position optimization problems, the present invention can help determine the global optimal sensor position, achieve lower cost-effectiveness, and have higher fault tolerance.

[0386] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. A person of ordinary skill in the art will appreciate that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVDROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0387] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A navigation method for an intelligent navigation robot based on multi-sensor fusion, characterized in that: The following steps are involved: S1, obtaining environmental data: using lidar, visual sensors and ultrasonic sensors to simultaneously collect environmental information; S2, data preprocessing: noise reduction, calibration and synchronization of the acquired sensor data; S3, environmental modeling: building an environmental model based on fused sensor data; S4, path planning: Generate the optimal navigation path using the environment model and path planning algorithm; S5, path planning: Generate the optimal navigation path using the environment model and path planning algorithm.

2. The intelligent navigation robot navigation method based on multi-sensor fusion as claimed in claim 1, characterized in that: The data preprocessing step further comprises: Filter the lidar data to remove noise points; Perform image enhancement and feature extraction on visual sensor data; Perform distance correction on ultrasonic sensor data to eliminate measurement errors.

3. The intelligent navigation robot navigation method based on multi-sensor fusion as claimed in claim 1, characterized in that: The environmental modeling step includes: Use lidar data to build a 2D map of the environment; Obstacle identification and labeling through image data obtained by visual sensors; The three-dimensional environment model is constructed by integrating the two-dimensional map and obstacle information.

4. The navigation method of an intelligent navigation robot based on multi-sensor fusion as claimed in claim 1, characterized in that: The path planning step includes: The dung beetle optimization algorithm is used for global path planning to generate a preliminary navigation path; The Dijkstra algorithm is used to locally optimize the preliminary path to ensure the smoothness and safety of the path.

5. The intelligent navigation robot navigation method based on multi-sensor fusion as claimed in claim 1, characterized in that: The real-time correction step comprises: The current position of the robot is estimated in real time through the particle filter algorithm; Adjust navigation paths based on real-time position and sensor feedback; An obstacle avoidance algorithm is used to replan the path when there are dynamic obstacles on the path.

6. An intelligent navigation robot based on multi-sensor fusion, characterized in that: include: A sensor suite that includes lidar, vision sensors, and ultrasonic sensors to simultaneously collect environmental information; A central processing unit that receives data from the sensor suite and uses it to perform tasks such as environment modeling, path planning, and real-time corrections; Motion control system, which drives the movement and steering of the robot according to the control instructions of the central processor; Power supply system, providing stable power supply for the robot; Communication module, used for communication between the robot and the remote control system or other robots.

7. The intelligent navigation robot based on multi-sensor fusion according to claim 6, characterized in that: The central processing unit further comprises: A data preprocessing module for noise reduction, calibration, and synchronization of data from the sensor suite; Environmental modeling module, which builds environmental models based on fused sensor data; The path planning module uses the environment model and path planning algorithm to generate the optimal navigation path; The real-time correction module performs real-time position correction and dynamic obstacle avoidance through sensor feedback to ensure that the robot navigates along the optimal path.

8. A robot path planning system based on dung beetle optimization algorithm, characterized in that: The system includes: The initial position setting module is used to initialize the initial position of the individuals of the dung beetle population by improving the Tent chaotic mapping; The behavior mode module sets two behavior modes, obstacle-free and obstacle-involved, according to the robot's motion environment in a two-dimensional static space, and uses a specific formula to update the position of the dung beetle; Adaptive perturbation module, used to introduce adaptive t-distribution perturbation in the process of dung beetle position update to enhance global search capability; The breeding behavior module is used to calculate the egg-laying area and location of dung beetles during reproduction; The foraging behavior module is used to determine the best foraging area and foraging location for dung beetles when foraging; A stealing behavior module, used to update the position of the dung beetle according to the stealing behavior of the dung beetle; The Cauchy mutation and reverse learning module is used to perturb and mutate the global optimal solution through Cauchy mutation and reverse learning strategies in the later stage of the algorithm; The path planning module constructs a grid map and performs path planning based on the robot's movement in a two-dimensional static space, using the path length as the design criterion of the fitness function.

9. The robot path planning system based on the dung beetle optimization algorithm according to system claim 8, further comprising: The obstacle detection module is used to define the obstacle-free environment as a free grid and the obstacle-containing environment as a prohibited grid when constructing a grid map; The fitness function calculation module is used to determine the starting point and end point of the robot's planned path, initialize the middle point of the path at the robot's current row, and determine the fitness function value based on whether there are obstacles around the middle point.

10. The robot path planning system based on the dung beetle optimization algorithm according to system claim 8, wherein: The fitness function calculation module also includes the following steps: Step 1. Initialize a path midpoint at the current robot row; Step 2. Determine whether the new path point will cause the robot to encounter an obstacle; If the new path point is an obstacle, the surrounding non-obstacle grids are selected as the new intermediate path point; If the new path point is not an obstacle and the point is not in the existing path, add it to the path; Step 3. Repeat steps 1 and 2 until the path connects the starting point and the end point, and calculate the total length of the path as the fitness function value.

Citation Information

Patent Citations

  • Processing method of operation map information

    CN108733060A

  • Robot and method for controlling movement of robot

    CN110053682A

  • Intelligent aged-supporting accompany robot

    CN110605724A

  • Three-dimensional map construction method, sweeping robot and electronic equipment

    CN111679663A

  • Intelligent wheelchair based on laser and visual SLAM fusion

    CN111708368A

Cited By

  • Flexible operation control system based on mechanical arm control algorithm

    CN119704193A

  • Underground space suitability evaluation method based on fuzzy hierarchical synthesis method and semi-supervised learning

    CN119917947A

  • Underground space suitability assessment method based on fuzzy hierarchical synthesis method and semi-supervised learning

    CN119917947B

  • Submersible data acquisition method and device based on deep learning

    CN120932083A

  • Deep learning based underwater vehicle data collection method and apparatus

    CN120932083B