Combined Motion Estimation and Tracking Control for Autonomous Navigation

摘要

This paper presents an assisting system for autonomous vehicle in outdoor environment. The system consists of several modules, which includes localization estimation, tracking control for navigation. In order to provide the position of a vehicle, which supports for future navigation, a motion estimation method based on nonholonomic constraint is presented. The method simplifies solution to 3D motion estimation using minimum set of parameters of geometric constraint. To reduce the number of parameters for accelerating computational speed, it is supposed that the vehicle moves following the model of Ackermann steering constraint, which requires constraints between rotation and translation components. An edge matching method based on omnidirectional vision is used to estimate the oriented heading of the vehicle motion. The advantage of the omnidirectional camera is that allows tracking landmarks in large rotation angle, which is utilized for high accuracy estimating. Finally, a stable and robust control method is used for motion tracking and control to navigate vehicle. The simulation and experimental results demonstrate the effectiveness in accuracy of the proposed method under variety of terrains in outdoor environments.