一种鲁棒的多目视觉惯性即时定位与建图方法  

A Robust Multi-Camera Visual-Inertial Simultaneous Localization and Mapping Method

作　　者：傅博 焦艳梅 丁夏清 吴俊[1,2] 熊蓉[1,2] FU Bo;JIAO Yanmei;DING Xiaqing;WU Jun;XIONG Rong(Institute of Cyber-Systems and Control,Zhejiang University,Hangzhou 310027,China;State Key Laboratory of Industrial Control Technology,Zhejiang University,Hangzhou 310027,China)

机构地区：[1]浙江大学智能系统与控制研究所,杭州310027 [2]浙江大学工业控制技术国家重点实验室,杭州310027

出　　处：《载人航天》2019年第5期581-585,共5页Manned Spaceflight

基　　金：国家重点研发计划(2017YFC0806501);省级重点研发计划(2019C01043)

摘　　要：针对深空环境下机器人精确自定位的大视野和防抖性要求,提出了一种基于多图像匹配和惯性约束的即时定位与建图(SLAM)方法。基于对多目情况下SLAM系统的像素投影与位姿跟踪过程的分析,引入紧耦合视觉惯性约束,以降低轨迹估计的累计漂移误差,增强系统的鲁棒性,使其在快速运动时定位不易丢失;针对鲁棒初始化的需求,提出了一种稳定的自动静止初始化方法。搭建了多目相机视觉惯性定位的仿真试验平台,在V-REP中搭建仿真试验场景,仿真结果验证了本方法的定位准确性和鲁棒性。To solve the long-term localization problem in the complex deep space exploration environment,a Simultaneous Localization and Mapping(SLAM)method based on multi-image matching and inertia constraint was proposed.The pixel projection and pose tracking process of the SLAM system were analyzed under the multi-camera condition and the tightly coupled visual inertia constraints were introduced to reduce the accumulated drift error of trajectory estimation and enhance the robustness of the system.Thus the localization was difficult to lose during fast motion.Moreover,a stable automatic static initialization method was proposed to realize the robust initialization.In the end,a simulation experiment platform for multi-camera visual-inertial localization was built in V-REP to verify the localization accuracy and robustness of the proposed method.

关 键 词：多目相机 视觉惯性紧耦合优化 自动静止初始化 即时定位与建图 

分 类 号：V448.2[航空宇航科学与技术—飞行器设计]