Development of a general purpose tour-guide robot able to learn routes from people and to adapt and move in unstructured and crowded environments

Abstract

Industrial robots has consolidated its presence in factories and professional environments during the last few years. However, in the next decades it is expected that service robots approach an exponential growth, reaching the mass market and taking over our homes and offices. In order to meet these predictions, there is still some way to go. Service robots need to be able to adapt to the dynamic nature of unstructured environments, where people walk by, and furniture might be moved around. In recent years, tour-guide robots have become a popular research topic because they face many of the challenges which have arisen in service robotics. Essentially, a tour-guide robot must be able to interact with humans, who might demand routes from the robot. Also, in order to showcase these routes, the robot must have a reliable and flexible perception of the changing environment where it operates. For these reasons, the goal of this thesis is to build a general purpose tourguide robot, which should be able to learn routes while following an instructor. The instructor can be anyone, i.e members of the staff from where the robot operates. This makes necessary the development of robust and flexible behaviours. More specifically, we have a) designed and developed a person-following behaviour, b) a human-robot interaction scheme, c) the processes for route recording and reproduction, and d) a strategy to allow the robot to learn its own control algorithms...