Intelligent autonomous obstacle avoidance and path optimization system for AI unmanned aerial vehicle

Abstract

The invention provides an intelligent autonomous obstacle avoidance and path optimization system for an AI unmanned aerial vehicle, and aims to realize efficient autonomous obstacle avoidance and optimal path planning of the unmanned aerial vehicle in a complex environment by integrating a plurality of sensor data fusion technologies, edge calculation processing and an advanced artificial intelligence algorithm. The system comprises a sensing module, a data processing module, an artificial intelligence decision module, an execution module and a communication module. Wherein the sensing module acquires environmental data in real time through various sensors such as a laser radar, a high-definition camera, an infrared sensor and an ultrasonic sensor; the data processing module performs data preprocessing and fusion by using edge computing equipment to generate a high-precision environment model; the artificial intelligence decision module comprises an obstacle avoidance sub-module and a path optimization sub-module which are respectively based on a deep learning algorithm and a reinforcement learning algorithm to realize identification and avoidance of dynamic and static obstacles and optimal path planning; the execution module controls the flight attitude, speed and course of the unmanned aerial vehicle according to the decision instruction; the communication module realizes real-time data communication between the unmanned aerial vehicle and a ground control station or other unmanned aerial vehicles, and supports cooperative flight and task allocation. According to the system, through combination of multi-sensor fusion and an advanced algorithm, the flight safety of the unmanned aerial vehicle is improved, the task execution efficiency is remarkably improved, and the system has wide application prospects and remarkable technical advantages.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles (UAVs), and in particular to an intelligent autonomous obstacle avoidance and path optimization system for unmanned aerial vehicles (UAVs) based on artificial intelligence (AI). The system integrates multiple sensor data fusion technologies, edge computing processing, and advanced artificial intelligence algorithms to achieve efficient autonomous obstacle avoidance and optimal path planning for unmanned aerial vehicles in complex environments, thereby improving flight safety and mission execution efficiency. Background Art

[0002] With the rapid development of science and technology, drones have been widely used in many fields such as logistics distribution, environmental monitoring, agricultural spraying, military reconnaissance, etc. However, in actual applications, drones often need to cross complex and changing environments when performing tasks, such as urban high-rise areas, densely forested areas, and rugged mountainous terrain. There are a large number of dynamic and static obstacles in these environments, such as buildings, wires, trees, other drones or birds in flight, etc.

[0003] Traditional obstacle avoidance and path planning methods for drones mainly rely on pre-set flight paths or real-time obstacle avoidance based on simple sensor data. These methods perform well in simple or known environments, but in complex and dynamically changing environments, they often face problems such as poor real-time performance, inaccurate obstacle avoidance, and inefficient path planning, making it difficult to meet the needs of efficient and safe flight.

[0004] In recent years, the rapid development of artificial intelligence technology, especially deep learning and reinforcement learning, has provided new solutions for autonomous obstacle avoidance and path optimization of drones. By using AI technology, drones can more intelligently perceive the environment, predict the movement trajectory of obstacles, and adjust the flight path in real time to achieve more efficient and safer flight. However, there are still many challenges in current research and application, such as the real-time fusion of multi-sensor data, the stability of algorithms in complex environments, and the overall integration of the system.

[0005] Therefore, there is an urgent need for an intelligent autonomous obstacle avoidance and path optimization system for drones that integrates multi-sensor perception, edge data processing, and advanced AI algorithms to overcome the shortcomings of existing technologies and improve the application performance of drones in complex environments. Summary of the invention

[0006] The purpose of this invention is to provide an AI UAV intelligent autonomous obstacle avoidance and path optimization system, which integrates multiple sensor data fusion technologies, edge computing processing and advanced artificial intelligence algorithms to achieve efficient autonomous obstacle avoidance and optimal path planning for UAVs in complex environments, thereby improving flight safety and mission execution efficiency.

[0007] To achieve the above-mentioned purpose, the present invention provides an AI UAV intelligent autonomous obstacle avoidance and path optimization system, comprising a perception module, a data processing module, an artificial intelligence decision module, an execution module and a communication module, wherein: the perception module comprises a laser radar, a high-definition camera, an infrared sensor and an ultrasonic sensor, which are used to collect three-dimensional data and dynamic information of the UAV's surrounding environment in real time;

[0008] The data processing module is used to pre-process and fuse the multi-source sensor data from the perception module to generate a high-precision real-time environment model;

[0009] The artificial intelligence decision-making module includes an obstacle avoidance submodule and a path optimization submodule. The obstacle avoidance submodule realizes the recognition and avoidance of dynamic and static obstacles based on a deep learning algorithm, and the path optimization submodule uses reinforcement learning or genetic algorithm to perform optimal path planning;

[0010] The execution module is used to control the flight control system of the UAV according to the instructions of the artificial intelligence decision module, including attitude adjustment, speed control and heading change;

[0011] The communication module is used to realize real-time data communication between the UAV and the ground control station or other UAVs, and supports coordinated flight and task allocation.

[0012] Preferably, the laser radar of the perception module is installed at the top center of the drone, high-definition cameras are distributed in the front and sides of the drone, infrared sensors are arranged at the front and rear ends of the drone, and ultrasonic sensors are installed at the bottom of the drone for detecting ground obstacles.

[0013] Preferably, the data processing module uses Kalman filtering and particle filtering algorithms to perform multi-sensor data fusion to generate a unified environmental perception model.

[0014] Preferably, the obstacle avoidance submodule of the artificial intelligence decision-making module uses a YOLO convolutional neural network to detect and classify obstacles, and uses an LSTM recurrent neural network to predict the motion trajectory of dynamic obstacles.

[0015] Preferably, the path optimization submodule of the artificial intelligence decision-making module adopts a deep Q network (DQN) for path planning, and combines a genetic algorithm to perform a global search for path optimization; the execution module includes a posture control unit, a speed control unit and a heading control unit, which are respectively used to control the posture, speed and heading of the drone; the communication module supports the 5G wireless communication protocol, which can achieve high-speed, low-latency data transmission, ensuring the real-time and stability of the system.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] Improve flight safety: By real-time identification and prediction of the position and movement trajectory of dynamic and static obstacles, the risk of collision can be effectively avoided.

[0018] Improve mission execution efficiency: Through the intelligent path optimization algorithm, the optimal flight path can be dynamically planned according to mission requirements, shortening flight time and reducing energy consumption.

[0019] Enhance the real-time performance and stability of the system: Use edge computing devices for data preprocessing and fusion to improve the system's response speed and data processing capabilities, ensuring the stability of the drone during high-speed movement.

[0020] Support multi-UAV collaborative operation: Realize real-time data sharing and collaborative control between UAVs through the communication module, and improve the efficiency and reliability of overall mission execution.

[0021] Adapt to complex and changing flight environments: The system can adaptively adjust the flight path in various complex and dynamically changing environments and has a wide range of application adaptability. DETAILED DESCRIPTION

[0022] To further illustrate the implementation of the present invention, each module of the system is described in detail below in conjunction with specific embodiments.

[0023] The overall architecture of this system includes five parts: perception module, data processing module, artificial intelligence decision module, execution module and communication module. Each module is interconnected through a high-speed data bus and wireless communication interface to form an efficient and coordinated intelligent obstacle avoidance and path optimization system.

[0024] Perception Module

[0025] The perception module is the "eye" of the system, responsible for collecting information about the drone's surroundings in real time. To achieve high-precision environmental perception, this module is equipped with a variety of sensors, including but not limited to:

[0026] LiDAR: Provides high-resolution 3D point cloud data that can accurately measure the distance and shape of obstacles around the drone.

[0027] HD Camera: Captures high-quality image data for visual recognition and obstacle detection.

[0028] Infrared sensor: Provides additional temperature and motion information to enhance environmental awareness in low-light or nighttime environments.

[0029] Ultrasonic sensor: used for close-range obstacle detection and provides precise distance measurement.

[0030] The data from these sensors are transmitted to the data processing module via a high-speed data bus to ensure real-time performance and data integrity.

[0031] Data processing module

[0032] The data processing module is the data center of the system. Its main task is to pre-process and fuse the multi-source sensor data from the perception module to generate a high-precision real-time environment model. It includes the following steps:

[0033] Data preprocessing:

[0034] Filtering and noise reduction: Filter sensor data to remove noise and interference signals and improve data quality.

[0035] Data synchronization: Time synchronization of asynchronous data from different sensors to ensure data consistency and timing.

[0036] Data Fusion:

[0037] Multi-sensor fusion algorithm: Use fusion algorithms such as Kalman Filter and Particle Filter to fuse data from different sensors and generate a unified environmental perception model.

[0038] 3D environment modeling: Use the fused data to build a high-precision 3D environment model to accurately reflect the obstacles and terrain features around the drone.

[0039] Edge computing optimization:

[0040] Real-time processing: Utilize the powerful computing power of edge computing devices to achieve real-time processing of large amounts of sensor data and reduce data transmission delays.

[0041] Distributed computing: When multiple drones work together, a distributed computing architecture is used to improve the overall processing power and scalability of the system.

[0042] The environmental model generated by the data processing module will serve as the input of the artificial intelligence decision-making module for it to make decisions on obstacle avoidance and path optimization.

[0043] Artificial Intelligence Decision-Making Module

[0044] The AI ​​decision module is the "brain" of the system, responsible for making intelligent decisions based on the environmental model, including obstacle avoidance and path optimization. This module consists of two submodules:

[0045] The algorithm flow of the artificial intelligence decision-making module is as follows:

[0046] Obstacle avoidance module:

[0047] Obstacle detection and identification --> dynamic obstacle prediction --> collision risk assessment --> obstacle avoidance strategy generation

[0048] Path optimization submodule:

[0049] Mission goal definition --> path planning algorithm selection --> path optimization --> dynamic adjustment

[0050] Obstacle avoidance module

[0051] The obstacle avoidance module is based on a deep learning algorithm to identify and avoid dynamic and static obstacles. Its main functions include:

[0052] Obstacle detection and identification:

[0053] Target detection algorithm: Deep learning models such as Convolutional Neural Network (CNN) are used to analyze sensor data, detect and classify surrounding obstacles. Specific algorithms such as YOLO (You Only Look Once) and SSD (Single Shot MultiBox Detector) can achieve real-time and efficient target detection.

[0054] Dynamic obstacle prediction: Use time series data and motion models to predict the future position and trajectory of dynamic obstacles. Use models such as Recurrent Neural Network (RNN) or Long Short-Term Memory (LSTM) to perform time series prediction.

[0055] Real-time obstacle avoidance decision-making:

[0056] Collision risk assessment: Based on the position and movement trajectory of obstacles, potential collision risks are assessed. The collision probability is quantified by calculating parameters such as the distance and relative speed between the drone and the obstacle.

[0057] Obstacle avoidance strategy generation: Use rule-based decision trees or reinforcement learning algorithms to generate obstacle avoidance strategies and adjust the flight path and posture of the drone to avoid collisions. Reinforcement learning models such as Deep Q-Network (DQN) and Policy Gradient Methods can learn the optimal obstacle avoidance strategy in a dynamic environment.

[0058] Path optimization submodule

[0059] The path optimization submodule uses optimization algorithms such as reinforcement learning, genetic algorithm or ant colony algorithm to dynamically plan the optimal flight path according to the mission objectives. Its main functions include:

[0060] Mission objectives definition:

[0061] Shortest time: Under the premise of ensuring safety, plan the shortest flight time path to improve the efficiency of mission execution.

[0062] Minimum energy consumption: Optimize flight paths to reduce energy consumption and extend the drone's endurance.

[0063] Minimum path length: Plan the shortest path distance to improve task execution efficiency.

[0064] Path planning algorithm:

[0065] Reinforcement Learning: Through interactive learning with the environment, the flight strategy is gradually optimized to achieve optimal path planning. Deep Reinforcement Learning (DRL) methods such as Deep Q Network (DQN) and Deep Deterministic Policy Gradient (DDPG) are used.

[0066] Genetic algorithm: Use the evolutionary mechanism of genetic algorithm to search the optimal path solution space and find the path solution that meets the task objectives. Through operations such as selection, crossover, and mutation, the path is gradually optimized.

[0067] Ant Colony Algorithm: It simulates the foraging behavior of ants and searches for the optimal path through the accumulation and volatilization of pheromones. It is suitable for complex path search and optimization problems.

[0068] Dynamic adjustment and optimization:

[0069] Real-time path adjustment: Dynamically adjust the flight path according to environmental changes and mission requirements to ensure the optimality and safety of the path. Adapt to changes in dynamic environments through online learning and real-time optimization.

[0070] Multi-objective optimization: Under the premise of meeting multiple task objectives, comprehensive optimization is carried out to balance factors such as time, energy consumption and path length. Multi-objective optimization algorithms such as NSGA-II (Non-dominated Sorting Genetic Algorithm II) are used to achieve coordination among multiple objectives.

[0071] Execution Module

[0072] The execution module is responsible for converting the instructions output by the artificial intelligence decision module into specific flight actions of the drone. Its main functions include:

[0073] Flight Control:

[0074] Attitude adjustment: Control the roll, pitch and yaw of the drone to adjust the flight attitude to meet the needs of path planning. Accurate attitude adjustment can be achieved through attitude control algorithms such as PID control (Proportional-Integral-Derivative Control) and fuzzy control.

[0075] Speed ​​control: According to the path optimization results, adjust the flight speed to achieve the best flight efficiency. Through the speed control algorithm, the flight speed of the drone is dynamically adjusted to adapt to different flight environments and mission requirements.

[0076] Heading change: Control the heading of the drone to ensure that it flies according to the planned path. Through the heading control algorithm, the heading angle of the drone is adjusted in real time to maintain the accuracy of the flight path.

[0077] Drive system interface:

[0078] Motor control: By adjusting the motor speed, the drone can move up and down, forward and backward. Motor control algorithms such as vector control and direct torque control (DTC) are used to ensure the rapidity and stability of the motor response.

[0079] Servo control: Control the servo movement to adjust the drone's attitude and direction. Through the servo control algorithm, precise servo movement control is achieved to ensure the stability of the drone's attitude.

[0080] Feedback Mechanism:

[0081] Real-time feedback: Monitor flight status and feed actual flight data back to the decision-making module for closed-loop control. Sensors are used to monitor the attitude, speed, position and other parameters of the drone in real time to ensure the accuracy of flight control.

[0082] Fault detection and processing: Detect possible faults during flight and take appropriate emergency measures to ensure flight safety. Real-time monitoring and fault processing of the flight system are achieved through fault detection algorithms, such as redundancy detection and anomaly detection.

[0083] Communication Module

[0084] The communication module enables real-time data communication between the drone and the ground control station or other drones, supporting collaborative flight and task allocation. Its main functions include:

[0085] Data transmission:

[0086] High-bandwidth communication: Wireless communication technologies (such as 5G, Wi-Fi, dedicated radio, etc.) are used to achieve high-speed data transmission and ensure real-time information sharing between modules. Multi-band and multi-channel communication technologies are used to improve the stability and reliability of data transmission.

[0087] Low-latency communication: Optimize communication protocols, reduce data transmission delays, and improve the system's real-time response capabilities. Use real-time transmission protocols (such as UDP, MQTT, etc.) to ensure the immediacy of data transmission.

[0088] Multi-UAV collaboration:

[0089] Task allocation: When multiple drones work together, the communication module is used to allocate and coordinate tasks to avoid task conflicts. A distributed task allocation algorithm is used to achieve dynamic allocation and adjustment of tasks.

[0090] Information sharing: Realize the sharing of environmental information and status information between drones to improve the efficiency and reliability of overall mission execution. By sharing environmental models, flight status and other information, collaborative obstacle avoidance and path optimization can be achieved.

[0091] Ground control station interface:

[0092] Remote monitoring: Realize real-time connection between the drone and the ground control station, support flight status monitoring, mission command issuance and data feedback. Provide real-time flight status display and control operations through the graphical user interface (GUI).

[0093] Emergency control: In an emergency, the ground control station can intervene in time to perform manual control or issue emergency commands to ensure flight safety. Emergency communication mechanisms are used to ensure that commands are communicated and executed at critical moments.

[0094] System workflow diagram:

[0095] The perception module collects environmental data

[0096] |

[0097] vData processing module performs preprocessing and fusion

[0098] |

[0099] v Generate environmental model and input it into the artificial intelligence decision module

[0100] |

[0101] v Artificial intelligence decision-making module for obstacle avoidance and path optimization decisions

[0102] |

[0103] vExecution module receives instructions and controls flight

[0104] |

[0105] v Real-time feedback of flight status to decision module

[0106] |

[0107] vCommunication module realizes data transmission and coordination

[0108] The specific workflow is as follows:

[0109] Environmental perception: The perception module collects three-dimensional data and dynamic information of the drone's surrounding environment in real time through multiple sensors such as lidar, high-definition cameras, infrared sensors and ultrasonic sensors.

[0110] Data processing: The data processing module filters, reduces noise, synchronizes and fuses the collected multi-source sensor data to generate a high-precision three-dimensional environment model.

[0111] Decision making: The AI ​​decision module performs obstacle detection, dynamic prediction, risk assessment and path optimization based on the environmental model, and generates obstacle avoidance and path planning instructions.

[0112] Flight control: The execution module receives instructions from the decision module and adjusts the attitude, speed and heading of the drone to ensure safe flight along the planned path.

[0113] Real-time feedback: During the flight control process, the execution module monitors the flight status in real time and feeds back the actual flight data to the decision-making module for closed-loop control and dynamic adjustment.

[0114] Data communication: The communication module enables real-time data transmission between the UAV and the ground control station and other UAVs, supports collaborative flight and task allocation, and improves the efficiency and reliability of overall mission execution.

[0115] Example 1: Flying in a city high-rise area

[0116] In urban areas with tall buildings, drones need to pass through dense buildings and face complex obstacle avoidance and path planning challenges. Using the AI ​​drone intelligent autonomous obstacle avoidance and path optimization system of the present invention, the drone collects data about the surrounding environment in real time through lidar and high-definition cameras. The data processing module fuses multi-source data to generate an accurate three-dimensional environmental model. The artificial intelligence decision-making module identifies the location and movement trajectory of tall buildings through the obstacle avoidance submodule, and uses the path optimization submodule to plan the optimal flight path to avoid the narrow space between tall buildings and achieve safe and efficient flight.

[0117] Workflow:

[0118] LiDAR and cameras collect 3D point cloud and image data of high-rise buildings.

[0119] The data processing module performs filtering, noise reduction and data fusion to generate a high-precision three-dimensional environment model.

[0120] The obstacle avoidance module identifies tall buildings and predicts possible collision risks.

[0121] The path optimization submodule plans the optimal flight path to avoid obstacles between high-rise buildings.

[0122] The execution module adjusts the attitude and heading of the drone to fly according to the planned path.

[0123] The communication module transmits flight status to the ground control station in real time, supporting mission monitoring and adjustment.

[0124] Example 2: Flying in a densely forested area

[0125] In densely forested areas, drones need to cross a large number of trees and dynamic obstacles, such as flying birds. The system of the present invention uses laser radar, high-definition cameras and infrared sensors to collect the three-dimensional structure of trees and the position of dynamic obstacles in real time. The data processing module fuses multi-source data to generate a detailed three-dimensional environmental model. The artificial intelligence decision-making module identifies the position of trees and birds through the obstacle avoidance submodule, predicts the movement trajectory of birds, and uses the path optimization submodule to plan the optimal flight path to avoid trees and dynamic obstacles, thereby achieving safe and efficient flight.

[0126] Workflow:

[0127] LiDAR collects 3D point cloud data of trees, HD cameras capture images of birds, and infrared sensors detect temperature and movement information of birds.

[0128] The data processing module performs data fusion and generates a detailed three-dimensional environmental model including trees and birds.

[0129] The obstacle avoidance module identifies trees and birds and assesses the risk of collision.

[0130] The path optimization submodule plans the optimal flight path avoiding trees and birds.

[0131] The execution module adjusts the attitude and heading of the drone to fly according to the planned path.

[0132] The communication module transmits the flight status in real time, supporting the monitoring and command of the ground control station.

[0133] Example 3: Mountainous Rough Terrain Flight

[0134] In the rugged terrain of mountainous areas, drones need to cross mountains, canyons and irregular terrain. The system of the present invention collects three-dimensional data of the terrain in real time through laser radar and high-definition cameras. The data processing module fuses multi-source data to generate a detailed three-dimensional environmental model. The artificial intelligence decision module identifies the location of mountains and canyons through the obstacle avoidance submodule, assesses the risk of collision, and uses the path optimization submodule to plan the optimal flight path to avoid rugged terrain and achieve safe and efficient flight.

[0135] Workflow:

[0136] LiDAR collects 3D point cloud data of mountains and canyons, and high-definition cameras capture terrain images.

[0137] The data processing module performs filtering, noise reduction and data fusion to generate a detailed three-dimensional environment model.

[0138] The obstacle avoidance module identifies mountains and valleys and assesses potential collision risks.

[0139] The path optimization submodule plans the optimal flight path that avoids rough terrain.

[0140] The execution module adjusts the attitude and heading of the drone to fly according to the planned path.

[0141] The communication module transmits the flight status in real time, supporting the monitoring and command of the ground control station.

[0142] Part VI

[0143] Preferred Embodiments of the Invention

[0144] On the basis of the above embodiments, the present invention may also be combined with other technical means and optimization measures to further improve the performance and adaptability of the system.

[0145] Optimization measure 1: Multi-sensor data synchronization and delay compensation

[0146] In order to ensure the time synchronization of multi-source sensor data and avoid data errors caused by delays, the present invention adopts high-precision clock synchronization technology, such as GPS clock synchronization and network-based time synchronization protocols (such as IEEE 1588 Precision Time Protocol, PTP), to achieve accurate time alignment of sensor data. At the same time, a delay compensation algorithm is used to dynamically adjust the data processing order according to the transmission delay of sensor data to ensure data consistency and timing.

[0147] Optimization 2: Adaptive training of deep learning models

[0148] In order to improve the adaptability of obstacle avoidance and path optimization algorithms in different environments, the present invention adopts an adaptive training mechanism to continuously optimize the parameters of the deep learning model through online learning and continuous training. The specific implementation includes:

[0149] Online learning: During flight, new environmental data and obstacle avoidance cases are collected in real time to update the weights and parameters of the deep learning model.

[0150] Transfer learning: Apply pre-trained models to new environments and quickly adapt to different environmental characteristics through transfer learning technology.

[0151] Data enhancement: Through data enhancement techniques such as image rotation, scaling, and noise addition, the training data set is expanded to improve the generalization ability of the model.

[0152] Optimization measure 3: Energy management and optimization

[0153] In order to extend the flight time of the drone, the present invention adopts an intelligent energy management strategy to reduce energy consumption by optimizing the flight path and flight strategy. The specific implementation includes:

[0154] Dynamic energy consumption assessment: Real-time monitoring of the drone’s energy consumption, adjustment of strategies based on flight status and path, and optimization of energy usage.

[0155] Energy-saving path planning: During the path optimization process, priority is given to path plans with lower energy consumption, such as reducing sharp turns and lowering flight speed.

[0156] Energy recovery mechanism: During landing or deceleration, energy recovery technology is used, such as motor braking to feed back electrical energy, to improve energy utilization efficiency.

[0157] Optimization measure 4: Multi-UAV collaborative obstacle avoidance and path planning

[0158] When multiple drones work together, the present invention uses a communication module to achieve information sharing and collaborative control between drones, thereby improving the efficiency and reliability of overall obstacle avoidance and path planning. The specific implementation includes:

[0159] Collaborative obstacle avoidance: UAVs share their own environmental models and flight status, and make collaborative obstacle avoidance decisions to avoid mutual interference and collision.

[0160] Collaborative path planning: Through collaborative path planning algorithms, path coordination between drones is achieved to optimize overall flight efficiency and mission execution results.

[0161] Load balancing: When multiple drones work together, tasks and resources are dynamically allocated to ensure load balancing and efficient operation of each drone.

[0162] Optimization measure 5: Fault tolerance and self-healing mechanism

[0163] In order to improve the stability and reliability of the system, the present invention adopts fault tolerance and self-healing mechanism to ensure that when a fault occurs, the system can detect it in time and take corresponding emergency measures. The specific implementation includes:

[0164] Fault detection: Detect possible faults such as sensor failure, motor failure, etc. by monitoring flight status and sensor data in real time.

[0165] Fault diagnosis: Fault diagnosis algorithms, such as Bayesian networks and support vector machines (SVM), are used to classify and diagnose detected faults and determine the fault type and location.

[0166] Emergency handling: Take appropriate emergency measures based on the type of fault, such as switching to the backup system, performing an emergency landing, adjusting the flight path, etc., to ensure flight safety.

[0167] Self-healing mechanism: Through the self-healing algorithm, the system can automatically adjust the operating status and restore normal functions after a failure occurs, reducing human intervention and maintenance costs.

[0168] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An AI UAV intelligent autonomous obstacle avoidance and path optimization system, characterized in that: It includes a perception module, a data processing module, an artificial intelligence decision module, an execution module and a communication module, among which: The perception module includes a laser radar, a high-definition camera, an infrared sensor, and an ultrasonic sensor, which are used to collect three-dimensional data and dynamic information of the drone's surrounding environment in real time; The data processing module is used to pre-process and fuse the multi-source sensor data from the perception module to generate a high-precision real-time environment model; The artificial intelligence decision-making module includes an obstacle avoidance submodule and a path optimization submodule. The obstacle avoidance submodule realizes the recognition and avoidance of dynamic and static obstacles based on a deep learning algorithm, and the path optimization submodule uses reinforcement learning or genetic algorithm to perform optimal path planning; The execution module is used to control the flight control system of the UAV according to the instructions of the artificial intelligence decision module, including attitude adjustment, speed control and heading change; The communication module is used to realize real-time data communication between the UAV and the ground control station or other UAVs, and supports coordinated flight and task allocation.

2. The AI ​​UAV intelligent autonomous obstacle avoidance and path optimization system according to claim 1 is characterized in that: The laser radar of the perception module is installed at the top center of the drone, high-definition cameras are distributed in the front and sides of the drone, infrared sensors are arranged at the front and back ends of the drone, and ultrasonic sensors are installed at the bottom of the drone for detecting ground obstacles.

3. The AI ​​UAV intelligent autonomous obstacle avoidance and path optimization system according to claim 1 is characterized in that: The data processing module uses Kalman filtering and particle filtering algorithms to fuse multi-sensor data and generate a unified environmental perception model.

4. The AI ​​UAV intelligent autonomous obstacle avoidance and path optimization system according to claim 1 is characterized in that: The obstacle avoidance submodule of the artificial intelligence decision-making module uses the YOLO convolutional neural network to detect and classify obstacles, and uses the LSTM recurrent neural network to predict the motion trajectory of dynamic obstacles.

5. The AI ​​UAV intelligent autonomous obstacle avoidance and path optimization system according to claim 1 is characterized in that: The path optimization submodule of the artificial intelligence decision-making module adopts deep Q network (DQN) for path planning, and combines genetic algorithm to perform global search for path optimization; the execution module includes attitude control unit, speed control unit and heading control unit, which are used to control the attitude, speed and heading of the drone respectively; the communication module supports 5G wireless communication protocol, which can realize high-speed and low-latency data transmission, ensuring the real-time and stability of the system.