In support of DARPA's Software-enabled Control project lead by Northrop Grumman (NG) Corporation, U.C. Berkeley is developing practical applications of a new control paradigm of hybrid systems for multi-vehicle, multi-modal control. The control design for hybrid systems needs to provide guarantees on safety, performance, fault tolerance, and mission completion in order to deliver high levels of mission reliability. Specifically, we will develop techniques for modeling, synthesis, and verification of control designs and their computational realization. We are working with Northrop Grumman to expand the hybrid control technology to meet design challenges provided in motivating scenarios for control of teams of autonomous unmanned air vehicles (UAVs). The outcome of this joint work will contribute to consolidation of the design specifications for the DARPA Open Control Platform (OCP).
The areas of research are:
Multi-vehicle Architecture Integration. Our approach in this project is to develop control design and implementation methodologies for both single vehicle and multiple vehicle control systems. We will be guided from the outset by our collaboration with NG to identify flight specifications for control UAVs, utilizing our group's experience designing control laws for the autonomous rotorcraft UAVs fabricated at Berkeley. We will address mode switching in the UAV control laws, degraded modes of operation, and multi-UAV coordination.
Multi-Modal Control Derivation and Analysis. This task forms the mathematical and control theoretical foundation of the hybrid systems-based approach. It will be driven by and fed back into the simulation results of the previous task. Key issues to be addressed in hybrid control design are numerical methods for optimal hybrid control, hierarchical hybrid control design, and designing for mission completion.
Safety and Performance Evaluation of Hybrid Control Designs. A control law evaluation environment and the underlying methodologies to cope with nonlinear scenarios, such as those which arise in a "swarm-of-UAVs," will be developed along the following lines: algorithmic analysis of nonlinear hybrid models, modular techniques for hybrid system validation, and model reduction and conservative approximation for hybrid models.
Hybrid System Simulation and Open Control Platform (OCP) Integration. This is aimed at taking the control design outcomes of the control system design tools that are developed in other thrusts and simulate and analyze them at a level closer to their software and hardware implementation. Our goal here is to estimate performance and validate safety of hybrid control designs (for flight applications and other safety critical operations) at the design level under consideration of implementation constraints such as hardware architecture, sensor fusion, resource sharing, and real time.
*This project is not officially supported through CITRIS funds, but the faculty and topical affiliations are sufficiently strong that it is listed here for referral and convenience.