Multiprocessing improvements on a low-cost system based Simultaneous Localization and Mapping

The process of Simultaneous Localization And Mapping (SLAM) has become a key topic of discussion amongst the research community in robotics and autonomous vehicle since its initiation in the latest decade. Many researchers consider SLAM ready to be instrumented for robot navigation but the implementation on a low-cost system has attracted few works. In this paper, an implementation of a SLAM algorithm is presented with regard to real-world operation. It is based on the co-design of a hardware architecture and an EKF-SLAM algorithm. Experiments were conducted with an instrumented vehicle in an a priori unknown environment. Results aim to demonstrate that our approach, based on low-cost sensors interfaced to an adequate multiprocessor architecture and an optimized algorithm, is good suitable to design embedded systems for SLAM applications in real world conditions.