基于AI技术的固定翼无人机集群分布式规避控制系统研究与设计

Research and Design of Distributed Avoidance Control System for Fixed Wing Drone Cluster Based on AI Technology

摘要:

固定翼无人机集群是由多个固定翼无人机组成的群体,在飞行任务中协同工作。相比于单个无人机,固定翼无人机集群具有更高的综合性能和更广阔的应用前景。但由于固定翼无人机在恶劣环境作业过程中存在着很多风险,例如障碍物撞击、相邻无人机碰撞等,威胁无人机的安全性。故设计了基于AI技术的固定翼无人机集群分布式规避控制系统。系统硬件单元主要包括AI视觉传感器设计单元、激光雷达传感器设计单元、控制器设计单元与无线通信网络设计单元,通过上述硬件设备提供实时的环境信息,控制无人机飞行姿态,实现信息共享。软件模块基于AI技术解算无人机位置和姿态,计算障碍物位置实现形状感知,完成无人机集群分布式规避控制。通过硬件与软件的协同作业,实现了固定翼无人机集群的分布式规避控制。实验结果显示:应用设计系统获得的无人机位置感知结果、障碍物位置感知结果与实际位置相同,路径总偏离量数值为4m,充分证实了设计系统控制性能较佳。

Abstract:

A fixed wing drone cluster is a group of multiple fixed wing drones that work together during flight missions. Compared to individual drones, fixed wing drone clusters have higher comprehensive performance and broader application prospects. However, due to the many risks associated with fixed wing drones operating in harsh environments, such as obstacle collisions, adjacent drone collisions, etc., the safety of drones is threatened. Therefore, a distributed avoidance control system for fixed wing unmanned aerial vehicle clusters based on AI technology was designed. The system hardware unit mainly includes an AI visual sensor design unit, a LiDAR sensor design unit, a controller design unit, and a wireless communication network design unit. Through the above hardware devices, real-time environmental information is provided to control the drone"s flight attitude and achieve information sharing. The software module is based on AI technology to calculate the position and attitude of drones, calculate the position of obstacles to achieve shape perception, and complete distributed avoidance control of drone clusters. Through the collaborative operation of hardware and software, distributed avoidance control of fixed wing drone clusters has been achieved. The experimental results show that the position perception results of the unmanned aerial vehicle and obstacle obtained by the application design system are consistent with the actual position, and the total deviation value of the path is 4m, fully confirming that the design system has better control performance.