Modeling Human Steering Behavior in Haptic Shared Control of Autonomy-Enabled Unmanned Ground Vehicles

Area of Science:

• Robotics
• Human-Computer Interaction
• Autonomous Systems

Background:

• Existing human steering models are limited to full teleoperation, not shared control with autonomy.
• Understanding human interaction with autonomous systems in Unmanned Ground Vehicles (UGVs) is crucial for effective collaboration.

Purpose of the Study:

• To extend human steering models for haptic shared control in autonomy-enabled UGVs.
• To mathematically capture human operator responses to autonomy and model human-autonomy collaboration.

Main Methods:

• Human subject tests were conducted to gather data for model development and validation.
• The ACT-R architecture and a two-point steering model were adapted to predict operator steering angles.
• A torque conversion module was developed to translate model commands into haptic feedback, and a parameterization strategy optimized performance based on Average Lane Keeping Error (ALKE).

Main Results:

• The developed model successfully predicts the minimum Average Lane Keeping Error (ALKE) achieved by human subjects in a shared control setting.

Conclusions:

• The extended human steering model accurately predicts optimal haptic shared control performance, measured by ALKE.
• This model facilitates simulation-based engineering for developing and evaluating shared control technologies in UGVs.