Scan Matching Cross-Correlation-Based Localization Algorithm: Embedded Systems Implementation Perspective

摘要

Localization in an indoor environment is a challenging aspect of an autonomous mobile robot control as such mobile systems typically have limited computational capacity. It is therefore essential to design an algorithm capable of reliable robot’s position calculation and, at the same time, suitable for implementation in embedded devices. This contribution describes a modified point-to-point Scan Matching algorithm based on cross-correlation, which was first introduced in our previous study. The newly developed algorithm running in a single thread is implemented in C language and decreases the computational complexity, i.e. time needed for calculation of the robot’s position, by as much as 20% compared to the original solution implemented in C#. LiDAR sensor was used to test the algorithm.