Smooth path planning with pedestrian avoidance for wheeled robots: Implementation and evaluation

Abstract

We studied smooth path planning using predesigned steering sets including three trajectory types. Path planning with a steering set is used to generate trajectories with smooth changes in direction. To put our path planning to practical use in dynamic environments, it needs to detect the movement of humans and to avoid moving humans. It is important that robot motion is fast and efficient, while also remaining safe for nearby humans. Robots need to find paths in human environments and to consider human movement for avoiding moving humans. Assuming that we can predict the trajectories of moving obstacles, we extend our method of smooth path planning to work in the presence of moving obstacles. We also define a new path evaluation method suitable for wheeled robots and experimentally evaluate our planner in an office environment, confirming the efficiency of our planning using our evaluation method.