Towards a Mobile-Robot Following Controller Using Behavioral Cues

This paper describes work towards a mobile-robot following controller which has the ability to incorporate a leader's behavioral cues into its controller formulation. The paper presents the mathematical formulation of the controller, and presents robot experimental studies used to investigate the controller. The controller continuously estimates the future predicted position of the leader (robot or human) as he/she/it moves, and then directs the follower robot to this position. A Kalman filter is employed for estimation that uses vision-based measurements of leader position, a dynamics model of the leader, and a behavioral-cue model of the leader. Singer's model is used to propagate the leader's state. A behavioral-cue model serves to create pseudo-measurements to further help the Kalman filter estimate the leader's future position. The controller is implemented on an ER Scorpion robot. Experiments are conducted using several different controllers. Results demonstrate that compared to other controllers, the proposed controller can more consistently follow the leader around sharp corners where line-of-sight is lost, as can happen often in indoor environments. However, in cases of more gradual movement where line-of-sight is not lost, a simpler vision-only controller has advantages.