Supporting Play by Applying Haptic Guidance Along a Surface Learnt from Single Motion Trajectories

Area of Science:

• Robotics
• Human-Computer Interaction
• Assistive Technology

Background:

• Haptic-enabled teleoperated robots offer potential for children with physical disabilities to interact with objects.
• Current assistive technologies may not fully compensate for limitations in reaching and object manipulation.

Purpose of the Study:

• To preliminarily test a learning from demonstration (LfD) approach for a haptic guidance system.
• To model a guidance surface using a neural network based on demonstrated game trajectories.
• To identify necessary improvements before trials with children with physical disabilities.

Main Methods:

• Adult participants demonstrated a whack-a-mole game using teleoperation.
• A neural network modeled a surface approximating demonstrated motion trajectories.
• Participants played the game with and without induced robotic movement errors requiring haptic guidance.

Main Results:

• Participants exhibited increased time, longer robot movement distances, and jerkier movements when using haptic guidance compared to no guidance.
• The system successfully learned a surface representing optimal game interaction trajectories.
• The study highlighted areas for system enhancement for future pediatric trials.

Conclusions:

• Learning a guidance surface, rather than a single trajectory, is a key contribution for adaptable robotic assistance.
• The developed LfD approach shows promise for creating intuitive robotic systems for children with physical disabilities.
• Further refinement is needed to optimize the haptic guidance for improved user experience and performance.