Trajectory generation for wheeled mobile robot based on landmarks

This paper describes a trajectory generation method for wheeled mobile robot. When a vehicle autonomously moves in an unknown environment, the vehicle must make the motion planning in real time. If a rough guideline is given for the trajectory, the motion planning becomes easy relative to the situation without any landmarks. However, even in that case, the vehicle must generate a desired trajectory for itself from the information on the guideline, for example, the positional data of the guideline obtained by using a vision system mounted on the vehicle. The authors propose a new trajectory generation method using Bezier curves, which is suitable for real time processing and has the capability of smoothly reducing the deviation from the guideline. In this paper, the trajectory generation method is described in detail, then the effectiveness is confirmed by simulations and experiments using a wheeled mobile robot developed by the authors.