A navigation system based on vision and motion fusion information using two UFKs

Abstract

This paper proposes an autonomous navigation system for mobile robots using two unscented Kalman filters (UKFs) and a slip detector to fuse the vision and motion information. The vision information is extracted from images captured by cameras, while the motion data are gathered by two wheel encoders, an accelerometer and a gyroscope. Firstly, the description of navigation algorithm, including the system overview, the image processing procedure and the coordinate transformation, are presented. Then kinematic models of two UKFs and data integration are introduced. Analyzing the results of experiments, the multi-sensor fusion system has more stability and accuracy in comparison with the single sensor system.