基于改进视觉惯性里程计的无人机系统设计  

UAV system design based on an improved visual inertial odometry

作　　者：王瑞 王茂森[1] 戴劲松[1] 管红根[1] WANG Rui;WANG Maosen;DAI Jinsong;GUAN Honggen(School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China)

机构地区：[1]南京理工大学机械工程学院,南京210094

出　　处：《兵器装备工程学报》2023年第4期275-281,共7页Journal of Ordnance Equipment Engineering

摘　　要：对小型旋翼无人机在缺乏GNSS信号情况下的定位导航问题,提出了基于改进视觉惯性里程计的无人机软硬件方案。首先设计带优先级的样本点采集策略和并行化计算方法,改进传统RANSAC算法,嵌入VINS-Fusion并用于无人机定位;其次,搭建无人机硬件平台,设计了飞控通信接口与控制系统。同时为高效地验证算法有效性,借助gazebo实现了无人机系统仿真。实验结果表明:算法在基础矩阵计算中平均内点比例提高1.8%,使用算法改进后的里程计在MH_01序列上的均方根误差降低0.001 m,在MH_02序列上的均方根误差降低0.019 m,满足无人机定位与导航时对精度与实时性的要求。Aiming at the positioning and navigation of small rotary-wing UAVs in the absence of GNSS signals,this paper proposes a UAV software and hardware solution based on an improved visual inertial odometry.Firstly,the sampling point acquisition strategy and the parallel calculation method with priority are designed,the traditional RANSAC algorithm is improved,and VINS-Fusion is embedded and used for UAV positioning.Secondly,the UAV hardware platform is built,and the flight control communication interface and control system is designed.At the same time,in order to effectively verify the validity of the algorithm,the simulation of the UAV system is realized with the help of gazebo.The experimental results show that the average interior point ratio of the algorithm in the basic matrix calculation increases by 1.8%,the root mean square error of the odometry after using the improved algorithm on the MH_01 sequence reduces by 0.001 m,and the root mean square error of the MH_02 sequence reduces by 0.019 m,which meets the accuracy and real-time requirements of UAV positioning and navigation.

关 键 词：视觉惯性里程计 改进RANSAC 外点剔除 VINS-Fusion 无人机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]