Influence of wheel-terrain contact point information errors on the accuracy of three-dimensional space odometry using tactile wheels and gyros

Abstract

A three-dimensional space odometry (3D-odometry) method that combines odometer data with gyroscope data is developed for self-localization of mobile robots on uneven terrains. However, the method incurred accumulated errors due to the lack of terrain information. In our previous study, we proposed a new 3D-odometry to increase the accuracy of the method. The improved 3D-odometry utilized wheel-terrain contact point information, which was detected by tactile wheels. Therefore, the accuracy of this method relied on the accuracy of the wheel-terrain contact point information. In this paper, the influence of errors of wheel-terrain contact point information was investigated by utilizing a self-localization simulation of a bike robot traveling on uneven terrain.