摘要
近年来视觉惯性SLAM是机器人及计算机视觉领域中的研究热点问题,在无人驾驶、增强现实、三维重建等领域有着广泛的应用。简述了视觉惯性SLAM的关键技术,并从统一的贝叶斯框架出发介绍了后端的两种形式——滤波和优化。对几个代表性视觉惯性SLAM系统进行深入对比分析,讨论了近年研究热点与发展趋势,并做出总结与展望。
Visual inertial SLAM is a popular research topic in both robotics and computer visual field in recent years.It has been widely used in the fields of Autonomous driving,Augmented reality and 3D reconstruction,etc.This paper describes the key techniques of visual inertial SLAM,and introduces 2 forms of backends filtering-based and optimization-based approaches from the unified Bayesian framework.Then,several representative visual inertial SLAM systems are deeply analysed and compared.Finally,some research hotspots and tendencies in recent years are discussed and summary and outlook is done.
作者
郭金辉
陈秀万
王媛
GUO Jin-hui;CHEN Xiu-wan;WANG Yuan(Institute of Remote Sensing and Geographic Information System,Peking University,Beijing 100871,China;China Academy of Electronics and Information Technology,Beijing 100041,China)
出处
《火力与指挥控制》
CSCD
北大核心
2021年第1期1-8,共8页
Fire Control & Command Control
基金
国家重点研发计划基金资助项目(2017YFC1500900,2018YFC0407702)。