Research And Design of A 3D Reconstruction Algorithm with Lio Algorithm

Abstract

Based on existing research results, this article uses the WheelTev Ackermann-type robot to implement the Fast-Lio and related improved algorithms and initializes system parameters under the condition of tight coupling of laser radar and inertial navigation unit data. At the same time, the obtained point cloud map is subjected to backend loop detection to remove graphic distortion generated during mapping and achieve 3D structure reproduction in multiple semi-open environments. Selected scenes with structural features in the campus of Southeast University were used to test the algorithm's effectiveness, and the robot was able to successfully complete mapping requirements and achieve automatic cruising and return in industrial scenes, demonstrating the algorithm's reliability in industrial scenes. The algorithm was also tested on some open-source datasets, and the experimental results showed improvements in some performance parameters, especially in computational load and robustness compared to existing algorithms.