Research
Our vision
My group's research aims to create “a new science for autonomy” by developing languages, theory, and algorithms for the design and verification of autonomous systems. This new science is in the making, and it is hard to predict what it will look like in the future. It is easier to predict what it will not be. For example, it will not fit within the current disciplinary boundaries. A number of disciplines, including control theory, formal methods, and learning, have a lot to offer for addressing the challenges in autonomy. Yet, an autonomous system is neither merely a physical object nor merely a piece of software nor merely a learning algorithm. The operation of an autonomous system relies on a harmonious integration of all these aspects and others.
The factors that drive our work
We take a relatively broad view on autonomy and tend to tackle abstract problems motivated by a collection of challenges cutting across multiple applications of autonomy.
Sample problems from our recent work
Toward living with autonomous systems we can trust
| I believe that, while improving the functionality and safety of autonomous systems, it is time to shape the future in which they are integrated into our lives at scale. Let me refer to a recent community-driven report on assured autonomy, sponsored by the Computing Community Consortium and several working groups at the Networking and Information Technology Research and Development Program. It argues: “Assurance, in the context of autonomy, cannot be an afterthought, and is it not just one of the qualifications of autonomy. Autonomy can survive as a useful technology only with proper assurance.” Furthermore, the report places the management of autonomous systems at the crossing of science, technology, society, policy, and governance. It points to the need for interdisciplinary approaches in order to realize the vast socio-economical opportunities and to address the associated challenges in autonomy. I believe that my group's philosophy for research is fully aligned with such an interdisciplinary approach and will help us define the future of research in autonomy. | 
|
Projects
Ongoing
- Autonomous Aerial Cargo Operations at Scale (NASA ULI)
- Cognitive Autonomy for Human CPS: Turning Novices into Experts (NSF, CPS)
- Testbed for Autonomy in Contested Environments (AFOSR, DURIP)
- Distributed Autonomous Information Acquisition and Management in LEO Satellite Networks (DARPA)
- Neuro-Symbolic Reinforcement Learning Under Perceptual Limitations (ONR)
- NASA Autonomy Verification and Validation Roadmap and Vision 2045 (NASA)
- Privacy-Performance Trade-Offs in Sequential Decision-Making (ONR)
- Modularity, Constraints and Multimodality in Learning for Sequential Decision-Making (ARO)
- Verification and Testing for Learning-Enabled Off-Road Autonomy (AFC)
- Hierarchical and Compositional Strategies for Control and Protection in PEPDS (ONR)
- Autonomous Learning with Probability and Abstraction for Competency Awareness (DARPA)
- Center of Excellence: Assured Autonomy in Contested Environments (AFOSR)
- Control, Learning and Adaptation in Information-Constrained, Adversarial Environments (DARPA)
- Verifiable, Control-Oriented Learning On The Fly (AFOSR, MURI)
- Joint Perception and Temporal Logic Planning for Distributed Agents in Dynamic Environments (ONR)
- Data-Driven Cyberphysical Systems (NSF, CPS)
- Co-Synthesis for Self-Awareness and Reconfiguration in Networked Systems (ONR)
- Safety-Constrained and Efficient Learning for Resilient Autonomous Space Systems (NASA)
- CAREER: Provably Correct Shared Control for Human-Embedded Autonomous Systems (NSF)
