Assistive front-following control of an intelligent robotic rollator based on a modified dynamic window planner

In this paper we present an assistive controller for the front-following task, applied to an intelligent robotic rollator. The control is based on a novel Dynamic Window local planner, which simultaneously treats discrete and continuous planning, by providing path equivalence classes and obstacle-free paths. This enables the detection of undecidability areas, e.g. junctions, as well as the safe locomotion of the robot while following the human. User experiments show that the assistive module, compared to a direct kinematic approach, allows the user to walk more naturally and more similar to his/her normal gait, and reduces the cognitive load.