Open AccessA Real-time Method on Robot with Continuous Trajectory for Low-drift Odometry and Mapping
Author(s) -
Jian Gao,
Lei Sun,
Xiaojia Xiang,
Zhenyi Li
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1576/1/012058
Subject(s) - odometry , trajectory , computer science , solver , simultaneous localization and mapping , artificial intelligence , position (finance) , computer vision , kinematics , mean squared error , range (aeronautics) , robot , control theory (sociology) , mathematics , mobile robot , engineering , physics , statistics , control (management) , finance , classical mechanics , astronomy , programming language , economics , aerospace engineering
This paper mainly proposes a real-time method on the robot with a continuous trajectory for low-drift odometry and mapping, by using range measurements from a 3D laser scanner, but without any other external reference. The common problem of the classical simultaneous localization and mapping (SLAM) method based on lidar is that the estimation of position and attitude is prone to jump discontinuity. It is mainly caused by the vulnerable performance of the optimization solver while it is in a complex scene. By adding kinematic and dynamic constraints into optimization function, our method achieves a lower mean square error(MSE) compare with the original one, and it has been tested in an outdoor experiment with only 3.5% MSE in yaw relative to state of the art.