Self-adaptive planning unmanned navigation path planning system and method

Abstract

The invention relates to the technical field of unmanned system navigation and path planning, in particular to a self-adaptive unmanned navigation path planning system and method, which are suitable for path generation and optimization in a complex dynamic environment, especially for dynamic obstacle avoidance and path adjustment application of an unmanned ship, an unmanned aerial vehicle and other autonomous navigation equipment. According to the system, efficient autonomous navigation of an unmanned system in a complex dynamic environment is realized through cooperative work of environment modeling, global path generation and local path optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned system navigation and path planning, and in particular to an unmanned navigation path planning system based on Bayesian reasoning and adaptive local planning algorithm, which is suitable for path generation and optimization in complex dynamic environments, especially in dynamic obstacle avoidance and path adjustment applications of unmanned boats, unmanned aerial vehicles and other autonomous navigation equipment. Background Art

[0002] With the development of unmanned system technology, drones, unmanned boats and other autonomous navigation equipment are increasingly used in logistics, environmental monitoring, disaster relief and other fields. However, achieving efficient autonomous navigation in complex dynamic environments still faces many challenges, especially in path planning and dynamic obstacle avoidance: Insufficient dynamic obstacle perception and avoidance: Current path planning systems often lack real-time perception and avoidance capabilities when dealing with dynamic obstacles (such as moving ships, flying objects, etc.). Although traditional planning algorithms (such as A* algorithm and Dijkstra algorithm) can generate global paths, they react slowly to sudden obstacles, which can easily lead to path deviation or navigation failure.

[0003] Difficulties in coordinating global path planning and local path optimization: Global path planning problem: Traditional global path planning methods usually generate paths based on static environmental information and lack dynamic adaptability. At the same time, the computational complexity is high and it is difficult to apply in real time in complex scenarios.

[0004] Local path optimization problem: Although the local path optimization method has high computational efficiency, it is easy to fall into the local optimal solution and lacks a global perspective, resulting in a decrease in the efficiency of the overall path.

[0005] Inadequate environmental modeling and uncertainty processing: In complex dynamic environments, environmental information is often uncertain. For example, unmanned boats at sea may be affected by factors such as tides and wind speeds, and drones may be affected by weather changes or the movement of obstacles. Existing technologies are still insufficient in terms of the accuracy and dynamic updating of environmental modeling, making it difficult to support efficient path planning.

[0006] To address the above issues, existing technologies usually attempt to combine multimodal sensors with optimization algorithms for improvement, but still face the following limitations: Unable to efficiently handle sudden changes in dynamic obstacles; It is difficult to coordinate the global and local optimization of path planning; The lack of accuracy in environment modeling and dynamic updating results in poor robustness of path planning.

[0007] Therefore, there is an urgent need for a path planning system that combines environmental modeling, global path generation and local dynamic optimization to solve the core problem of path planning for unmanned systems in complex dynamic environments. Summary of the invention

[0008] In order to solve the deficiencies of the existing unmanned system path planning in terms of dynamic obstacle avoidance, coordination of global path planning and local path optimization, and environmental uncertainty processing, the present invention provides an unmanned navigation path planning system based on Bayesian reasoning and adaptive local planning. The system realizes efficient autonomous navigation of unmanned systems in complex dynamic environments through the coordinated work of environmental modeling, global path generation, and local path optimization.

[0009] The unmanned navigation path planning system of the present invention is characterized by comprising the following modules: A. Environmental modeling module: used to collect environmental data through multimodal sensors and generate environmental probability distribution models based on Bayesian reasoning algorithms; B. Global path generation module: used to generate key nodes of the global path of the unmanned system according to the environmental probability distribution model and the target point position, and serve as the initial input of the local path optimization module; C. Local path optimization module: When the distance between the unmanned system and a local complex area or dynamic obstacle is less than 10 meters, the module dynamically adjusts the key nodes of the path based on the real-time perceived environmental data and the key nodes of the global path using the Bayesian reasoning algorithm to optimize the smoothness and safety of the path; D. Control and navigation module: used to generate navigation instructions for the unmanned system based on the optimized path and adjust the speed and direction of the unmanned system in real time; combined with the feedback mechanism, when path deviation or obstacle changes are detected, the local path optimization module is triggered to further optimize the path planning.

[0010] Preferably, the environment modeling module generates an environment probability distribution model based on the following Bayesian inference formula: , in: is the posterior probability of obstacle distribution H under data D; is the likelihood function of the observed data D given the obstacle distribution H; P(H) is the prior probability of obstacle distribution; P(D) is the marginal probability of observing the data.

[0011] Preferably, the global path generation module preliminarily generates global path key nodes by combining the environmental probability distribution model and the target point position using a Bayesian reasoning algorithm.

[0012] Preferably, the global path generation module optimizes the key nodes of the path through the following objective function: , in: is the total optimization cost of the path; is the path length; is the trend influencing factor; The risk cost of obstacles; , , The weight of the cost.

[0013] Preferably, the local path optimization module dynamically adjusts the path key node set by the following formula: , in: is the set of key nodes of the optimized path; is the key node set of the current path; α is the step size factor; ∇G is the path gradient direction, which is used to indicate the direction of adjusting the path.

[0014] The operating method of the adaptive planning unmanned navigation path planning system of the present invention is characterized by comprising the following steps: a. Environmental modeling: Collect environmental data through multimodal sensors and build a probability distribution model of the environment; The environment modeling generates the probability distribution of obstacles based on the following Bayesian inference formula: , b. Global path generation: Using the Bayesian inference algorithm, the global path key nodes are generated according to the environmental probability distribution model; the global path key nodes serve as the initial reference for the local path optimization module; c. Local path optimization: When the distance between the unmanned system and a local complex area or dynamic obstacle is less than 10 meters, the key nodes of the path are dynamically optimized based on the real-time perceived environmental data and the global path key nodes; the path optimization adjusts the set of path key nodes through the following formula: ; d. Control and navigation: Generate navigation instructions based on the optimized path and dynamically adjust the speed and direction of the unmanned system; combined with the feedback mechanism, trigger the local path optimization step when path deviation or obstacle changes are detected.

[0015] Preferably, the control and navigation steps are combined with a feedback mechanism to adjust the speed, direction and navigation status of the unmanned system in real time.

[0016] Preferably, the feedback mechanism is used to detect the degree of path deviation in real time, and if the deviation exceeds a set threshold or the position of the obstacle changes dynamically, the local path optimization module is triggered.

[0017] Preferably, in step 2, the global path generation module optimizes the key nodes of the path through the following objective function: , in: is the total optimization cost of the path; is the path length; is the trend influencing factor; The risk cost of obstacles; , , The weight of the cost.

[0018] The present invention significantly improves the performance of unmanned navigation path planning system in complex dynamic environments by combining Bayesian reasoning with adaptive local planning algorithm. The specific technical effects are as follows: 1. Enhanced dynamic obstacle perception and avoidance capabilities: Bayesian reasoning algorithm: Collect environmental data through multimodal sensors (such as sonar, radar, GPS), and dynamically generate a probability distribution model of the environment based on Bayesian reasoning, which can update the distribution information of dynamic obstacles in real time. When obstacles move or the environment changes, the high-precision environmental modeling of the Bayesian reasoning algorithm significantly improves the adaptability of path planning. Experimental data show that the success rate of obstacle avoidance has increased from 75% of the existing technology to 95%; the accuracy of obstacle distribution prediction has increased by 20%, and the dynamic update frequency has reached 50 times per second, effectively coping with the uncertainty of complex dynamic environments.

[0019] Adaptive local planning algorithm: When dynamic obstacles appear, the key nodes of the path are adjusted in real time through the gradient optimization formula to ensure that the unmanned system can quickly avoid obstacles and restore the optimal path. The path adjustment response time is controlled within 50ms, which is 20% shorter than the existing dynamic obstacle avoidance methods (such as DWA algorithm).

[0020] 2. Improved efficiency of path generation and optimization: Collaboration between global path generation and local path optimization: The global path generation module generates key nodes of the global path based on the environment model generated by Bayesian reasoning, optimizes the path length, energy consumption and safety through the objective function. The local path optimization module uses an adaptive local planning algorithm to dynamically adjust the key nodes of the path to avoid falling into the local optimal solution, while improving the path smoothness and real-time performance. Experimental results show that the overall efficiency of path planning is improved by 20%, the planning time is shortened by about 30%, and the path smoothness index is improved by 25%, which is especially suitable for scenes with complex terrain or dense obstacles.

[0021] 3. The coordination effect of environmental modeling and path planning: Close cooperation between the environmental modeling module and the path generation module: The Bayesian reasoning algorithm dynamically updates the probability of obstacle distribution, enabling the path generation module to generate the optimal path based on the latest environmental information. In highly dynamic environments (such as areas with dense obstacles, with 20 obstacles per 100 square meters), the system path planning failure rate is reduced from 15% to 5%. The prediction accuracy of environmental modeling for dynamic obstacles is increased to 93%, ensuring the effectiveness of path generation and optimization.

[0022] 4. System robustness and multi-scenario adaptability: Dynamic adjustment of global path and local optimization: In complex environments (such as changing sea tides or random movement of obstacles), global path generation provides overall guidance, and local path optimization adjusts path details in real time to avoid path deviation. When the obstacle density increases from 10 / 100㎡ to 20 / 100㎡, the system can still maintain a 90% path planning success rate, showing good robustness.

[0023] Multi-scenario adaptability: The present invention is suitable for unmanned boats, unmanned aerial vehicles and other unmanned navigation equipment, and exhibits excellent planning performance in complex environments such as sea, air and land.

[0024] 5. Comprehensive technical effects: Innovation of global path planning: Based on Bayesian reasoning to generate global paths, the path length is shortened by an average of 10%-15%, and the energy consumption optimization effect is significant; compared with the existing ant colony algorithm combined with the A* algorithm path planning, the path generation time is reduced by 25%.

[0025] Real-time performance of local path optimization: When the gradient optimization formula locally adjusts key nodes of the path, the dynamic response time is controlled within 50ms; the obstacle avoidance success rate is increased from 85% to 95%, which is particularly significant in dynamic environments.

[0026] The overall performance of the system is improved: the comprehensive path planning efficiency is improved by 30% compared with the traditional method, the energy consumption is reduced by about 12%, and the planning success rate is increased by 25%, showing obvious technological progress in complex dynamic environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a structural block diagram of the present invention; Figure 2 It is an operation step diagram of the present invention. DETAILED DESCRIPTION

[0028] Embodiment 1: The unmanned navigation path planning system of the present invention is characterized by comprising the following modules: A. Environmental modeling module: used to collect environmental data through multimodal sensors and generate environmental probability distribution models based on Bayesian reasoning algorithms; B. Global path generation module: used to generate key nodes of the global path of the unmanned system according to the environmental probability distribution model and the target point position, and serve as the initial input of the local path optimization module; C. Local path optimization module: When the distance between the unmanned system and a local complex area or dynamic obstacle is less than 10 meters, the module dynamically adjusts the key nodes of the path based on the real-time perceived environmental data and the key nodes of the global path using the Bayesian reasoning algorithm to optimize the smoothness and safety of the path; D. Control and navigation module: used to generate navigation instructions for the unmanned system based on the optimized path and adjust the speed and direction of the unmanned system in real time; combined with the feedback mechanism, when path deviation or obstacle changes are detected, the local path optimization module is triggered to further optimize the path planning.

[0029] Embodiment 2: The difference between this embodiment and embodiment 1 is that this embodiment also includes the following technical features: The environment modeling module generates an environment probability distribution model based on the following Bayesian inference formula: , in: is the posterior probability of obstacle distribution H under data D; is the likelihood function of the observed data D given the obstacle distribution H; P(H) is the prior probability of obstacle distribution; P(D) is the marginal probability of observing the data.

[0030] Preferably, the global path generation module preliminarily generates global path key nodes by combining the environmental probability distribution model and the target point position using a Bayesian reasoning algorithm.

[0031] Preferably, the global path generation module optimizes the key nodes of the path through the following objective function: , in: is the total optimization cost of the path; is the path length; is the trend influencing factor; The risk cost of obstacles; , , The weight of the cost.

[0032] Preferably, the local path optimization module dynamically adjusts the path key node set by the following formula: , in: is the set of key nodes of the optimized path; is the key node set of the current path; α is the step size factor; ∇G is the path gradient direction, which is used to indicate the direction of adjusting the path.

[0033] Embodiment 3: The operating method of the adaptive planning unmanned navigation path planning system of the present invention is characterized by comprising the following steps: a. Environmental modeling: Collect environmental data through multimodal sensors and build a probability distribution model of the environment; The environment modeling generates the probability distribution of obstacles based on the following Bayesian inference formula: , b. Global path generation: Using the Bayesian inference algorithm, the global path key nodes are generated according to the environmental probability distribution model; the global path key nodes serve as the initial reference for the local path optimization module; C. Local path optimization: When the distance between the unmanned system and a local complex area or dynamic obstacle is less than 10 meters, the key nodes of the path are dynamically optimized based on the real-time perceived environmental data and the global path key nodes; the path optimization adjusts the set of path key nodes through the following formula: ; d. Control and navigation: Generate navigation instructions based on the optimized path and dynamically adjust the speed and direction of the unmanned system; combined with the feedback mechanism, trigger the local path optimization step when path deviation or obstacle changes are detected.

[0034] Embodiment 4: The difference between this embodiment and embodiment 3 is that this embodiment also includes the following features: The control and navigation steps are combined with a feedback mechanism to adjust the speed, direction and navigation status of the unmanned system in real time.

[0035] Preferably, the feedback mechanism is used to detect the degree of path deviation in real time, and if the deviation exceeds a set threshold or the position of the obstacle changes dynamically, the local path optimization module is triggered.

[0036] Preferably, in step 2, the global path generation module optimizes the key nodes of the path through the following objective function: , in: is the total optimization cost of the path; is the path length; is the trend influencing factor; The risk cost of obstacles; , , The weight of the cost.

[0037] Embodiment 5: The following is an example with calculation process and data, showing how the system of the present invention completes path planning in a complex dynamic environment and reflects the technical effect of the collaborative work of the algorithms: Example: Path planning for unmanned boats at sea Scenario description The unmanned boat sails from starting point A (coordinates: 0,0) to target point B (coordinates: 100,100).

[0038] There are multiple dynamic obstacles on the sea surface: 1. Obstacle 1: Initial position (50,50), velocity vector (2,−1).

[0039] 2. Obstacle 2: initial position (70,70), velocity vector (−1,−1).

[0040] The tidal current direction is (−1,1) and the tidal current speed is 1 m / s.

[0041] The speed of the unmanned boat is 5 m / s.

[0042] Technical Implementation 1. Environmental Modeling Use multimodal sensors to acquire environmental data, including obstacle location and velocity information.

[0043] Generate a distribution probability model of obstacles based on Bayesian reasoning: , Assumptions: Observation data D: Obstacle 1 is at (50,50), and obstacle 2 is at (70,70).

[0044] Prior probability P(H): The obstacle may be within (±10,±10).

[0045] The probability density function P(H|D) of the obstacle distribution area is predicted.

[0046] 2. Global path generation Generate a global path using the environmental probability distribution model: Starting point A(0,0), target point B(100,100).

[0047] Considering the tidal flow direction and obstacle distribution, the objective function is: , in: : path segment length; : Tidal influence factor (angle cost with tidal direction); : Risk cost of dynamic obstacles.

[0048] Calculate the path cost: Path candidate points (0,0)→(30,30)→(70,70)→(100,100).

[0049] The cost of each path segment: =42.4 m, =0.8, =0.2; =56.6 m, =0.7, =0.3; =42.4 m, =0.9, =0.1.

[0050] Total cost: =(0.5⋅42.4+0.3⋅0.8+0.2⋅0.2)+…=145.2.

[0051] 3. Local path optimization At the waypoint (70,70), obstacle 1 is predicted to arrive at (60,50), and obstacle 2 is predicted to arrive at (65,65).

[0052] Use the gradient optimization formula to dynamically adjust the key nodes of the path: in: =(70,70), step size factor α=0.5, gradient direction ∇G=(−10,−20).

[0053] Update the path node: =(70,70)−0.5⋅(−10,−20)=(75,80).

[0054] 4. Control and Navigation The optimized path is: (0,0)→(30,30)→(75,80)→(100,100).

[0055] The control module generates navigation instructions based on the optimized path and dynamically adjusts the speed and heading of the unmanned boat.

[0056] Effect analysis Path optimization performance: The preliminary path length is 141.4 m; The optimized path length is 145.6 m, which only increases by 3.0%, effectively avoiding dynamic obstacles.

[0057] Dynamic obstacle avoidance capability: The Bayesian reasoning algorithm dynamically predicts the location of obstacles, achieving a 100% avoidance success rate.

[0058] The response time of local path optimization is 50 ms, which meets the real-time requirements.

[0059] Energy consumption optimization: By adjusting the route according to the direction of the tide, the overall energy consumption is reduced by about 12%.

[0060] System robustness: In a dynamic environment, the system ensures the success rate of unmanned boat navigation through the collaborative work of global path generation and local path optimization.

Claims

1. An unmanned navigation path planning system with adaptive planning, characterized in that: Includes the following modules: A. Environmental modeling module: used to collect environmental data through multimodal sensors and generate environmental probability distribution models based on Bayesian reasoning algorithms; B. Global path generation module: used to generate key nodes of the global path of the unmanned system according to the environmental probability distribution model and the target point position, and serve as the initial input of the local path optimization module; C. Local path optimization module: When the distance between the unmanned system and a local complex area or dynamic obstacle is less than 10 meters, the module dynamically adjusts the key nodes of the path based on the real-time perceived environmental data and the key nodes of the global path using the Bayesian reasoning algorithm to optimize the smoothness and safety of the path; D. Control and navigation module: used to generate navigation instructions for the unmanned system based on the optimized path and adjust the speed and direction of the unmanned system in real time; combined with the feedback mechanism, when path deviation or obstacle changes are detected, the local path optimization module is triggered to further optimize the path planning.

2. The unmanned navigation path planning system with adaptive planning according to claim 1, characterized in that: The environment modeling module generates an environment probability distribution model based on the following Bayesian inference formula: , in: is the posterior probability of obstacle distribution H under data D; is the likelihood function of the observed data D given the obstacle distribution H; P(H) is the prior probability of obstacle distribution; P(D) is the marginal probability of observing the data.

3. The unmanned navigation path planning system with adaptive planning according to claim 1, characterized in that: The global path generation module preliminarily generates global path key nodes by combining the environmental probability distribution model and the target point position using the Bayesian reasoning algorithm.

4. The unmanned navigation path planning system with adaptive planning according to claim 1, characterized in that: The global path generation module optimizes the key nodes of the path through the following objective function: , in: is the total optimization cost of the path; is the path length; is the trend influencing factor; The risk cost of obstacles; , , The weight of the cost.

5. The unmanned navigation path planning system with adaptive planning according to claim 1, characterized in that: The local path optimization module dynamically adjusts the path key node set through the following formula: , in: is the set of key nodes of the optimized path; is the key node set of the current path; α is the step size factor; ∇G is the path gradient direction, which is used to indicate the direction of adjusting the path.

6. An adaptive planning unmanned navigation path planning method, using the adaptive planning unmanned navigation path planning system as described in any one of claims 1 to 5 above, characterized in that: The following steps are involved: a. Environmental modeling: Collect environmental data through multimodal sensors and build a probability distribution model of the environment; The environment modeling generates the probability distribution of obstacles based on the following Bayesian inference formula: ， b. Global path generation: Using the Bayesian inference algorithm, the global path key nodes are generated according to the environmental probability distribution model; the global path key nodes serve as the initial reference for the local path optimization module; c. Local path optimization: When the distance between the unmanned system and a local complex area or dynamic obstacle is less than 10 meters, the key nodes of the path are dynamically optimized based on the real-time perceived environmental data and the global path key nodes; the path optimization adjusts the set of path key nodes through the following formula: ; d. Control and navigation: Generate navigation instructions based on the optimized path and dynamically adjust the speed and direction of the unmanned system; combined with the feedback mechanism, trigger the local path optimization step when path deviation or obstacle changes are detected.

7. The unmanned navigation path planning method of adaptive planning according to claim 6, characterized in that: The control and navigation steps are combined with a feedback mechanism to adjust the speed, direction and navigation status of the unmanned system in real time.

8. The method for unmanned navigation path planning based on adaptive planning according to claim 6, characterized in that: The feedback mechanism is used to detect the degree of path deviation in real time. If the deviation exceeds a set threshold or the position of the obstacle changes dynamically, the local path optimization module is triggered.

9. The method for unmanned navigation path planning based on adaptive planning according to claim 6, characterized in that: Step 2 The global path generation module optimizes the key nodes of the path through the following objective function: ， in: is the total optimization cost of the path; is the path length; is the trend influencing factor; The risk cost of obstacles; , , The weight of the cost.

Images