摘要
鉴于现有的无人机路径规划方法难以兼顾路径质量和计算效率,提出了一种将扰动流体动态系统与深度神经网络相结合的自适应路径规划方法。首先,基于扰动流体算法仿真生成样本数据以解决样本数量不足的问题,并通过灰狼优化算法和滚动时域控制优化样本性能。然后,利用深度神经网络强大的学习能力,从样本数据中提取无人机与目标点、障碍物之间的相对位姿信息作为深度神经网络的输入,以扰动流体算法的参数作为深度神经网络输出端的特征提取,离线训练深度神经网络。之后,利用训练好的深度神经网络基于当前环境信息实时动态调整扰动流体参数。最后,通过仿真验证得知该方法具有较高的鲁棒性,规划的路径质量较高,且计算开销较小符合实时性要求,增强了无人机对环境的自适应能力。
In view of the existing path planning methods of UAV are difficult to consider both path quality and computational efficiency,this paper proposes an adaptive path planning method by combining interfered fluid dynamical system with deep neural network.Firstly,the interfered fluid algorithm is used to generate sample data,and the grey wolf optimizer and rolling horizon control are used to optimize the sample performance.Then,with the strong learning ability of the deep neural network,the relative pose information between UAV,destination and obstacles are extracted from the sample data as the input of deep neural network,and the interfered fluid parameters are used as the feature extraction of the output end to train the deep neural network offline.Next,the trained deep neural network is used to dynamically adjust the parameters of the interfered fluid algorithm based on the current environmental information.Finally,the simulation results show that the proposed method has high robustness,the quality of the planned path is high,and the calculation cost is small,which meets the real-time requirements.The proposed method enhances the adaptive ability of UAV to the environment.
作者
王延祥
王宏伦
吴健发
伦岳斌
WANG Yanxiang;WANG Honglun;WU Jianfa;LUN Yuebin(School of Automation Science and Electrical Engineering,Beihang University,Beijing 100191,China;Science and Technology on Aircraft Control Laboratory,Beihang University,Beijing 100191,China)
出处
《无人系统技术》
2020年第6期50-58,共9页
Unmanned Systems Technology
基金
国家自然科学基金(61673042)。
关键词
无人机
路径规划
扰动流体动态系统
深度神经网络
灰狼优化算法
滚动时域控制
Unmanned Aerial Vehicle(UAV)
Path Planning
Interfered Fluid Dynamical System
Neural Net⁃work
Grey Wolf Optimizer
Rolling Horizon Control
