Odometry and Inertial Sensor-based Localization of a Snake Robot

Abstract

Terrain adaptability gives snake robots an edge over wheeled mobile robots. Snake robot's applications expansion to disaster management is due to its limbless and modular design. Replicating the navigation and locomotion of a biological snake have been in the spotlight for many researchers. In this study, a robust localization algorithm of a snake robot is developed. A sensor fusion-based algorithm using probabilistic approaches: UKF (Unscented Kalman Filter) and EKF (Extended Kalman Filter) has been proposed. Odometry is fused with one and two IMUs (Inertial Measurement Units) with both probabilistic approaches. Evaluation of results in a simulation environment showed that the fusion of two IMUs and odometry using EKF outer performs in terms of accuracy when odometry is fused with one IMU using EKF. Furthermore, the fusion of two IMUs and odometry with UKF is computationally expensive resulting in a large convergence time which is not the best suitable approach that can be utilized for the snake robot.