On the Initialization Problem for Timed-Elastic Bands

Abstract

Path planning is an important part of navigation for mobile robots. Several approaches have been proposed in the literature based on a discretisation of the map, including A*, Theta*, and RRT*. While these approaches have been widely adopted also in real applications, they tend to generate non-smooth paths, which can be difficult to follow, based on the kinematic and dynamic constraints of the robot. Time-Elastic-Bands (TEB) have also been used in the literature, to deform an original path in real-time to produce a smoother path, and to handle potential local changes in the environment, such as the detection of an unknown obstacle. This work analyses the effects on the overall path for different choices of initial paths fed to TEB. In particular, the produced paths are compared in terms of total distance, curvature, and variation in the desired heading. The optimised version of the solution produced by Theta* shows the highest performance among the considered methods and metrics, and we show that it can be successfully followed by an autonomous bicycle.