Hybrid feedback path following for robotic walkers via bang-bang control actions

This paper presents a control algorithm conceived to guide a robotic walking assistant along a planned path using electromechanical brakes. The device has been modelled as a Dubins' car, a wheeled vehicle that moves only forward in the plane and constrained by a lower bounded turning radius. In order to reduce the hardware requirements as much as possible, so as to reduce the costs, we work with limited sensory information, so the classic modulated braking guidance cannot be implemented. As a consequence, the vehicle has a very limited set fixed turning radius manoeuvres ruled by a hybrid controller based on three discrete states: go straight, turn right and turn left. Despite of these limitations and in order to ensure a gentle and natural convergence of the user to the planned path, the robot approaches the path with an orientation that depends on its position and that guarantees a continuous curvature of the tracked path. Results are validated using extensive simulations.