Invited Speakers

NeBula: Resilient autonomy under uncertainty

Ali Agha, Jet Propulsion Laboratory

Abstract: In this presentation, we will discuss JPL’s NeBula Autonomy and AI framework. NeBula aims at providing a holistic solution for single- and multi-robot systems autonomously operating in unknown environments. It is architected as a modular software package addressing stochasticity and uncertainty in various elements of the mission, including sensing, environment, motion, system health, and communication. We discuss NeBula’s performance on various robotic platforms (wheeled, legged, tracked, and flying vehicles) in the context of terrestrial and planetary-analog missions. Specifically, we discuss how a NeBula-powered team of robots has won Phase II of the DARPA Subterranean Challenge and has enabled the first fully autonomous exploration of unknown planetary-analog caves for 100s of meters. Finally, we will briefly discuss NeBula’s key modules, including 1) traversability, 2) estimation, 3) 3D mapping, 4) multi-robot coordination, 5) comm-aware formation, and 6) machine learning-based model adaptation. For more details on NeBula, please see the project’s website: https://costar.jpl.nasa.gov/ or our paper here: https://arxiv.org/abs/2103.11470

Bio: Dr. Agha’s research focuses on autonomy for robotic systems and spacecraft. Dr. Agha is the group leader for JPL’s Aerial Mobility Group. He also leads the Team CoSTAR and the development of the NeBula autonomy solution at JPL, which won the 2020 DARPA Challenge: Subterranean Urban Challenge. His team also developed autonomous multi-robot solutions, which placed second at the 2019 DARPA Challenge, Tunnel competition. At JPL, Dr. Agha is leading several robotic autonomy projects with a dual focus on planetary exploration and terrestrial applications. Previously, he was with Qualcomm Research, leading the perception efforts for self-flying drones and autonomous cars. Prior to that, Dr. Agha was a Postdoctoral Researcher at MIT. His research interests include artificial intelligence, autonomous decision making, and perception for robotic systems, with applications to legged robots, rovers, drones, and self-driving offroad vehicles. Dr. Agha was named NASA NIAC fellow in 2018.