EAGER: TaskDCL: Building Human Trust in Autonomous Social Navigation With Egocentric Visual Feedback

This EArly-concept Grant for Exploratory Research (EAGER) project aims to advance our understanding of customizing safe control strategies for Autonomous Mobility Systems (AMS) interacting with humans, thereby fostering scientific progress and national prosperity. Safety, crucial for AMS operating alongside people in environments like autonomous cars, wheelchairs, and delivery carts, varies among individuals, influencing their perception of safety risks. Misalignment between onboard safety protocols and human safety sensitivities can erode trust in AMS capabilities. This award supports fundamental research to enhance trust in AMS by using human egocentric eyeglasses, with the frame of reference defined from the users perspective, as external AMS sensors. This approach dynamically creates personalized safety models, adjusting AMS movements to align with estimated sensitivities. By better addressing user safety preferences, AMS technologies can gain broader public acceptance and trust. The societal impact extends across various sectors, ensuring efficient, safe transport in human-shared environments—from logistics to healthcare, enhancing mobility independence. This research also promotes inclusivity by involving underrepresented groups and fosters collaboration between egocentric vision and robotics communities. AMS requires a choice of safety model and associated parameters to ensure the plans and controls executed will avoid collision with obstacles, such as other humans. However, each individual interacting with the AMS has different safety preferences. This EArly-concept Grant for Exploratory Research (EAGER) project will consider (i) a computer vision model to identify safety risks in the environment based on an individual’s eye-gaze information, (ii) an online inference approach using egocentric visual feedback to estimate the individual’s safety preference, and (iii) a technique to synthesize empathetically astute AMS motions that adapt to those preferences, such as proactively yielding to accommodate a more cautious individual. The research team will demonstrate the efficacy of the approach on an autonomous wheelchair platform and evaluate the connection between empathetic astute motions and trust in AMS capabilities. This EAGER award has been co-funded by the Dynamics, Controls, and System Diagnostics and the Mind, Machine, and Motor Nexus Programs. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.