Delft Haptics Lab

The Effect of Online Adaptation on Conflicts in Haptic Shared Control for Free-Air Teleoperation Tasks

Abstract

This MSc project was performed by Arnold de Jonge, under supervision of Jeroen, Henri and David. His thesis led to a journal paper in Transactions on Haptics. For the Haptics Symposium 2016, this journal paper was selected as one of four papers to be presented in a special slot in the symposium; which Tricia Gibo did.

Abstract Haptic shared control provides artificial guiding forces on a control interface that support operators to perform tasks. Research has shown that this can improve task performance and reduce control effort. However conflicts between the human operator and support system are often reported which deteriorate performance and increase control effort. Conflicts solutions have been proposed for unexpected avoidance of obstacle on the supported trajectory, but solving conflicts due to trajectory negotiation mismatches have not been proposed. One way of minimizing those conflicts is to base support on individual preferred trajectories. Another way propose in literature is by online adapting the supporting trajectory.
The goal of this study is to provide evidence for the hypotheses that 1) an individualized support trajectory yields less conflict between operator and support system than convention general trajectory and 2) online adaptation of the supported trajectory will also reduce conflicts, regardless of the initially chosen support trajectory.
In a human factors experiment, subjects (n=12) conducted a repetitive two degrees of freedom task in which they were provided with four different types of support. Both the recorded individual trajectories and the conventional centerline of the environment trajectory were provided both with and without adaptation.
The results show no effect of adaptation or support path on performance. Adaptation reduces support forces and control activity. The individual recorded trajectory reduces control activity compared to conventional centerline of environment support.
Those results provide evidence that recorded individual trajectory support reduces control effort for non-adaptive support, albeit only in control activity. The results furthermore provide evidence that adaptation of both types of initial support path reduce support forces and control effort. In conclusion, online adaptation of haptic shared control reduces trajectory negotiation conflicts and the associated increased forces. It adapts to subject-specific preferences in trajectory, regardless of initially chosen supported trajectories.