A Parallel RatSlam C++ Library Implementation

Abstract

RatSlam is a bio-inspired Simultaneous Location and Mapping (SLAM) algorithm used for autonomous mobile robots navigation tasks. This work presents a RatSlam algorithm implementation as a C++ library designed to take advantage of internal RatSlam modules parallelization. The RatSlam algorithm is presented with principal aspects of the library architecture design. Furthermore, its results using a well known RatSlam data set with a standard RatSlam implementation (OpenRatSLAM - Robot Operating System), and a Python implementation. The mapping found with the previous approaches and the proposed on this work were similar. Moreover, the execution times between the OpenRatSLAM and this C++ library was compared, with the proposed implementation having a lower execution time. Thus, the current implementation was validated and has some advantages against previous ones, which can be very relevant for real-time applications.