紀要論文

差動二輪ロボットのためのSLAMアルゴリズム開発

DEVELOPMENT OF A NEW SLAM ALGORITHM DEVELOPMENT OF A NEW SLAM ALGORITHM

jpn

24368083

https://doi.org/10.15002/00030781

法政大学大学院理工学研究科

法政大学大学院紀要. 理工学研究科編

65

1

4

2024-03-24

Version of Record

In this paper, we describe the development of a new SLAM algorithm for differential drive robots using 3D LiDAR. Traditional SLAM scan matching often employs ICP (Iterative Closest Point) matching, which evaluates point-to-point correspondences, or NDT (Normal Distributions Transform) matching, which evaluates point-to-distribution correspondences. However, these methods require repetitive optimization, which can lead to local optima and fail to calculate the correct position and orientation (matching). Therefore, this paper proposes a new SLAM algorithm that leverages the characteristics of differential drive robots by limiting their movements to in-situ turning and straight-line movement. This approach eliminates the issues associated with ICP and NDT matching and treats the problem as a linear estimation issue, enabling fast and stable position and orientation estimation. The effectiveness of the proposed new SLAM algorithm for differential drive mobile robots is evaluated using a 3D robot simulator.

Article