Abstract

This research introduces a novel framework for safe human-robot interaction that combines active human state estimation, conformal trajectory prediction, and Model Predictive Control. The framework enables robots to maintain probabilistic safety guarantees while adapting to human internal states such as attention levels and driving styles. Through simulations of diverse driving scenarios, we developed a reward-based human behavior model. This approach advances human-robot interaction by providing an adaptive method for safe robot planning that explicitly accounts for human behavioral uncertainties.

Document Type

Article

Author's School

McKelvey School of Engineering

Author's Department

Electrical and Systems Engineering

Class Name

Electrical and Systems Engineering Undergraduate Research

Language

English (en)

Date of Submission

12-4-2024

Share

COinS