Position Estimation for Mobile Robots in Dynamic Environments

Abstract

For mobile robots to be successful, they have to navigate safely in populated and dynamic environments. While recent research has led to a variety of localization methods that can track robots well in {em static} environments, we still lack methods that can robustly localize mobile robots in dynamic environments, where, for example, people may block the robot's sensors for extensive periods of time. This paper proposes a family of probabilistic algorithms that can localize mobile robots even in densely populated environments. These algorithms are based on Markov localization, which estimates the location of a robot probabilistically. A novel entropy-based filter is employed for determining the "believability" of a sensor reading, thereby filtering out sensor readings that are corrupted by humans or unexpected changes in the environment. The technique was recently implemented and applied as part of an installation, in which a mobile robot gave interactive tours to visitors of the ``Deutsches Museum Bonn.'' Extensive empirical tests involving datasets recorded during peak traffic hours in the museum demonstrate that this approach is able to accurately estimate the robot's position in more than 99% of the cases even in such highly dynamic environments.