Hybrid SLAM-based Exploration of a Mobile Robot for 3D Scenario Reconstruction and Autonomous Navigation

Abstract

SLAM can be categorized into two groups: laser-based SLAM and visual-based SLAM. They are used to identify surrounded objects of a robot. This paper proposes a combination of visual-based SLAM algorithm and laser-based SLAM. The purpose is to reduce effort but still provides the high quality 3D-reconstructed map. First, this paper presents visual-based SLAM and laser-based SLAM separately. Then, two techniques are integrated into one system. In addition, bi-direction RRT* path planning algorithm is developed to create a feasible and optimal trajectory. A self-tuning Fuzzy-PID controller also is introduced for driving the robot to follow the trajectory precisely. The simulations and real experiments are conducted in order to illustrate the superiority of the proposed approach.