Project Title: High Confidence Reconfigurable Distributed Control
Organization: California Institute of Technology
Project URL: http://www.cds.caltech.edu/sec/
Objective: The objective of this research is to develop and validate a theoretical framework for high-confidence, multi-vehicle coordination and distributed control. We are developing an optimization-based, hierarchical control architecture for motion control systems and using modeling and analysis to explore trajectory spaces for aggressive flight vehicles. Experimental implementation is being done using the Caltech Ducted Fan, a simple flight control experiment that replicates the longitudinal dynamics of a thrust vectored aircraft.
Approach: We are using receding horizon, optimization-based control to allow online control customization for aggressive flight vehicles. This approach relies on the use of new theoretical results to provide guaranteed stability and near optimal control in a real-time environment. Proper consideration of constraints is a central element of the research effort, allowing exploration of maneuvers for flight vehicles that push the limits of performance. A unique feature of our work is the use of geometric methods for drastically reducing the computational requirements and allowing real-time implementation.
Recent Accomplishments: Implemented a real-time algorithm for trajectory generation in the presence of state and input constraints, based on combining ideas from differential flatness and collocation to obtain high performance algorithms. This approach has been demonstrated on simulations of the Caltech ducted fan and timing results indicate they can be implemented on the experimental system.

Derived new theoretical results that demonstrate that incremental improvements during iterations of the algorithm are sufficient to guarantee stability for receding horizon control techniques. This provides a solid framework for implementing algorithms that use a finite number of iterations at each step of the algorithm.

Developed a graphical technique for exploring operability maps for the ducted fan, allowing better understanding of the maneuvers that can be achieved for this aircraft.

Current Plan: Short term tasks for this project (00Q3-00Q4) are to develop improved models for the ducted fan, implemented in the Modelica modeling language, implement the current real-time trajectory generation algorithm in dSPACE, and building a hardware in the loop capability at Colorado for rapid prototyping and multi-vehicle simulation. We anticipate having the ability to demonstrate highly aggressive maneuvers using online, optimization-based control by October 2000. FY01 tasks are to develop optimization-based, hierarchical control laws that allow team oriented tasks, such transition from individual vehicle attitude control to coordinated wingtip tracking (formation flight), development of tools for exploring operability maps and flight envelopes, and beginning to build algorithms that guarantee high confidence operation of flocks of vehicles in the presence of faults and changes in mission.
Technology Transition: None completed. Working on transition of basic optimization-based control algorithms to UMN (Balas) and UTRC (Fuller) for work on this program.