UC San Diego

Robots have the potential to assist people in dynamic, safety-critical environments like disaster zones or emergency departments (EDs). They can reduce people's workload assisting people with low-risk tasks or provide people with a physical presence in remote or dangerous places. However, there are many obstacles to introducing robots into dynamic environments, including integrating them into these chaotic spaces without causing further disruption, ensuring they do not add to people's workload, and enabling people to easily control them.

The goal of my work is to develop robots that support users in dynamic, safety-critical environments. In this dissertation, I describe four main contributions of my work.

First, I explored ways to design new proximate human-robot teaming methods. I developed a fully autonomous robotic system that teamed with a person to engage in a series of key proximate teaming tasks: co-manipulation, handovers, and shared workspace tasks. Additionally, I presented a framework to aid researchers in designing proximate human-robot teaming studies.

Second, I explored how to design and situate robots in Emergency Medicine (EM) environments. I developed Iris, a low-cost telemedicine robot. Iris is an open-hardware, open-software robot that enables HCWs to visit patients remotely. I evaluated its usability and feasibility in the ED with HCWs. Based on their feedback, I developed concrete design suggestions for building and deploying telemedical systems for EDs, including ensuring situational awareness, accessibility, and familiarity and trust. This work lays the foundation for developing and deploying more usable, robust, and accessible robotic systems into ED environments.

Third, I investigated how mobile telemanipulator robots (MTRs) can support HCWs during interruptions. EM HCWs are often required to manage multiple tasks simultaneously and are frequently interrupted, so robots in these spaces must be developed with these interruptions in mind. I designed interruption-mitigation and reorientation methods for a MTR with insights drawn from the literature and in collaboration with EM physicians. I conducted a study to explore the efficacy of these strategies with EM physicians in a realistic simulation environment. Our investigations revealed how MTRs might support multitasking in environments with frequent interruptions, as well as practical and ethical challenges surrounding the integration of MTRs in the ED. Findings from this work will support the development of interruption- mitigation strategies to enable robots to better support people in fast-paced, interruption-driven environments.

Finally, I developed WISARD, a shared autonomy system that supports operators during periods of high workload. WISARD adapts the amount of assistance it provides based on the operator's workload in order to better support people in dynamic environments like EDs. I conducted an evaluation of WISARD, which elucidated factors that affect operators' perceptions and preferences, including sense of control and temporal efficiency. Additionally, my analysis revealed nuanced ways autonomy can support or hinder operators. These findings suggest that shared control systems need to be carefully personalized to each user. They also point to tensions between personalization and operator familiarity, as adapting systems might interfere with transparency and operator comfort. This work provides insights that HRI researchers can leverage to create MTR shared control systems that better support people during high workload periods.

My work contributes to the development of robotic systems that support and adapt to people in dynamic, safety-critical environments, which could decrease errors and improve task performance. My work informs the design of robots that can support people during periods of high workload. By considering the unique requirements of these spaces, my research enables robots to better assist people in their work.