Abstract

Mobile robots that are employed in people's homes need to safely navigate their environment. And natural human-inhabited environments still pose significant challenges for robots despite the impressive progress that has been achieved in the field of path planning and obstacle avoidance. These challenges mostly arise from the fact that (i) the perceptual abilities of a robot are limited, thus sometimes impeding its ability to see relevant obstacles (e.g. transparent objects), and (ii) the environment is highly dynamic being populated by humans. In this contribution we are making a case for an integrated solution to these challenges that builds upon the analysis and use of implicit human knowledge in path planning and a cascade of replanning approaches. We combine state of the art path planning and obstacle avoidance algorithms with the knowledge about how humans navigate in their very own environment. The approach results in a more robust and predictable navigation ability for domestic robots as is demonstrated in a number of experimental runs. ©2010 IEEE.