Intentional User Adaptation to Shared Control Assistance

Shared control approaches to robot assistance typically assume that user behavior remains the same despite the addition of the the assistance and rely upon this assumption to infer user goals. However, HRI research consistently shows that users are highly sensitive to changes in robot performance, which contradicts this assumption. In this paper, we show that users, in fact, change their control behavior in the presence of assistance and describe these changes as intentional adaptations to the new system dynamics. We present two user studies in which participants controlled robots with various levels of assistance. In a computer-based study, participants report changing their strategies as the assistance changes, and the amount of change in the direction of their control significantly differs between assistance conditions. In an in-the-wild robot study, participants teleoperated a robot to pick up a cup despite the presence of "assistance" that drives the system towards nonexistent goals and away from the true goals of the task. The ability of participants to overcome the assistance and still achieve the goal further demonstrates that users can change their behavior to account for the novel dynamics. These results motivate further research in user-centered design and evaluation of assistive systems that treat the user as intentional.