Center of Excellence: Assured Autonomy in Contested Environments

From u-t-autonomous.info
Jump to: navigation, search

A collaborative project supported by the Air Force Office of Scientific Research.

The Center's website, Oden Institute annoucement.

Collaborators:

  • University of Florida: Warren Dixon, John Shea, Matthew Hale, Norman, Norman Fitz-Coy, Kevin Butler
  • Duke University: Michael Zavlanos, Miroslav Pajic
  • University of California, Santa Cruz: Ricardo Sanfelice



The Center activities are in collaboration with the Air Force Office of Scientific Research (AFOSR) and the munitions (RW), sensors (RY), and space vehicles (RV) directorates within the Air Force Research Laboratory (AFRL).

Autonomous systems must execute high level missions plans with verifiable assurances despite uncertain adversarial environments where the integrity and availability of sensor information and communications are challenged. Key innovations include analysis, design and synthesis tools that enable autonomous mission execution despite uncertainty within complex dynamics while accounting for the integrity and privacy of information on computationally constrained resources.

To address such challenges, the members and collaborators of the Center will develop a tightly integrated set of fundamental theories and methods across six diverse research topics.

  • In Modeling and Analysis Methods for Nonsmooth Systems, we will develop a new modeling framework and analysis tools to examine dynamics that result from the integration of intermittent information with dynamic systems. Specifically, we develop analysis tools for nonsmooth switched and hybrid systems that account for various switching triggers and uncertainty. For complex nonlinear systems, design is inspired from analysis.
  • Hence, we develop adaptive and optimal design tools that are integrated with high-level logic-based control synthesis tools in Adaptation, Optimality, and Synthesis. The design and analysis tools provide a framework to incorporate different information flows within dynamic systems.
  • Since the union of physical dynamics and information dynamics are magnified within a multi-agent network, Analysis, Design and Control Synthesis Within and Over Networks outcomes enable scaled autonomous operations while considering the dynamics of the communications channel, including opportunities for adaptive network/controller interactions.
  • Design and Analysis with Asynchronous Information, develops tools for robustness against delays and asynchronous data exchanges within such networks. Information disruptions and delays can result from natural effects in the environment or be a design feature (e.g., to reduce bandwidth or power consumption); however, in contested environments, information perturbations can emanate from adversarial actors in unique and complex ways, requiring security-aware design and analysis methods.
  • In Attack-Resilient Designs, security techniques are developed for attack detection and resilient estimation and control for systems with constrained resources by exploiting environmental and agent dynamics and knowledge of expected agent interaction and sensing characteristics.
  • Adversaries can also engage in clandestine efforts where computations and communications can be covertly monitored/intercepted. Efforts in Protecting Safety- and Mission-Critical Information focus on obfuscating such efforts to ensure private actions by agents.