LCSR Seminar: Abriana Stewart-Height

Description

Abriana Stewart-Height, a postdoctoral fellow in the Department of Mechanical Engineering and a member of the Computer Science & Artificial Intelligence Laboratory at the Massachusetts Institute of Technology, will give a talk titled "People Helping Robots So That Robots Can Help People" for the Laboratory for Computational Sensing + Robotics.

Abstract:

As the rapidly changing climate leads to more extreme weather, the environment is becoming increasingly dangerous for people to conduct critical field operations such as fire suppression, ocean exploration, nuclear waste management, and disaster response. Autonomous robots that can operate in extreme environments—habitats characterized by harsh and dynamic conditions (e.g., temperature, terrain, obstacles)—would greatly reduce injuries of human workers in these conditions. Mobile robots have made great strides toward robust locomotion in unknown dynamic environments. Yet, these machines are not indestructible and increasing terrain complexity almost guarantees failure will occur. Using insights from deep learning, dynamical systems theory, and biology, I developed a unique data-driven approach to address fault detection and recovery in legged systems.

Critical field operations in extremely challenging environments often require more skill and knowledge than any singular person or machine can have. Shared autonomy holds great promise for applications ranging from space exploration to emergency medical response by distributing functions between humans and machines to accomplish a goal. More specifically, my recent work seeks to overcome uncertainty in object-based SLAM by enabling robots to learn to ask for help from other robots or humans. In this talk, I will first present my prior research that investigates resiliency in mobile robotic devices (e.g., legged robots and robotic walkers). Followed by my current research interests in long-term shared autonomy in complex and dynamic environments.