Person Following through Cluttered Environments using Probabilistic Methods and Fast Local Obstacle Avoidance

摘要

Person following through cluttered environments while avoiding obstacles is an important task for service robots. It provides, for example, a natural and simple way to guide the robot around in its new environment. The robot should be robust against distracting objects especially those, which are similar to human legs. Although an overwhelming amount of work is available to address the person detection problem, the challenge of the person following task itself has not been brought into focus very often. The problem of obstacle avoidance while following a person with robot-mounted sensors has to be solved. A method that helps avoiding blockades with obstacles while following through small and cluttered areas has to be found. This paper proposes a combined person tracking and motion planning approach. The combination of both techniques makes the person following a robust and usable method to guide a robot to everyday-life environments.