基于B样条的连续时间轨迹状态估计研究综述  

Review of Continuous-time Trajectory State Estimation Research Based on B-Splines

作　　者：吕佳俊 郎晓磊 李宝润 刘勇[1] LÜJiajun;LANG Xiaolei;LI Baorun;LIU Yong(College of Control Science and Engineering,Zhejiang University,Hangzhou 310058,China)

机构地区：[1]浙江大学控制科学与工程学院,浙江杭州310058

出　　处：《机器人》2024年第6期743-752,共10页Robot

基　　金：国家自然科学基金(62088101)。

摘　　要：多源数据融合是近年来状态估计技术的一大发展趋势,提高了状态估计的精度和鲁棒性。然而多传感器带来了许多新问题,如高频异频异步数据的时间域关联、传感器外参的准确标定、持续采集型传感器的数据畸变校正、异构传感器数据的融合等。连续时间轨迹方法在克服这些问题上具有天然的优势。本文对基于B样条的连续时间轨迹状态估计研究进行了综述。首先介绍基于B样条的连续时间轨迹状态估计理论,其次对离线标定和在线里程计的不同应用进行了分类梳理,最后展望了未来的研究发展方向。Multi-source data fusion is a major development trend in state estimation technology in recent years,enhancing the accuracy and robustness of state estimation.However,multi-sensor integration brings new challenges such as timedomain association of high-frequency,different-frequency,and asynchronous data,the accurate calibration of sensor extrinsic parameters,the data distortion correction of continuous acquisition sensors,and fusion of heterogeneous sensor data.Continuous-time trajectory methods naturally have advantages in overcoming these problems.This paper reviews the research on continuous-time trajectory state estimation based on B-splines.Firstly,the theory of continuous-time trajectory state estimation based on B-splines is introduced.Next,different applications to offline calibration and online odometry are systematically classified.Finally,future research directions are discussed.

关 键 词：连续时间轨迹 状态估计 B样条 传感器标定 同步定位与地图构建 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]