Large-scale monocular FastSLAM2.0 acceleration on an embedded heterogeneous architecture

Abstract

Simultaneous localization and mapping (SLAM) is widely used in many robotic applications and autonomous navigation. This paper presents a study of FastSLAM2.0 computational complexity based on a monocular vision system. The algorithm is intended to operate with many particles in a large-scale environment. FastSLAM2.0 was partitioned into functional blocks allowing a hardware software matching on a CPU-GPGPU-based SoC architecture. Performances in terms of processing time and localization accuracy were evaluated using a real indoor dataset. Results demonstrate that an optimized and efficient CPU-GPGPU partitioning allows performing accurate localization results and high-speed execution of a monocular FastSLAM2.0-based embedded system operating under real-time constraints.