LAUSR

Laser scanners have been verified successfully by academic research and industrial applications on their eligibility for robot localization tasks. This article introduces Circle Fit Matching (CFM), an analytical and efficient 2-D laser scan matching (LSM) algorithm for the motion estimation problem in structured environments. Unlike traditional iterative methods, we wrap the scan inputs with high-level planar information and model the two-scan transformation estimation problem in a pure geometric manner. The translation and rotation parts are decoupled and solved separately with the plane normal and plane-to-sensor distance. To avoid leading to unstable solutions, at least two nonparallel associated planes are necessary, then the degenerated cases are identified and processed to improve the accuracy of the laser odometry. Besides, a two-stage plane feature matching method is proposed. Experiments evaluating the proposed algorithm’s effectiveness, accuracy, and speed have been conducted in both simulation and real-world environments. Experimental results show that our algorithm is fast and effective for the planar environments. The implementation of our method is open source to benefit the robotics community and beyond http://github.com/bzdfzfer/circle_odometry.