Co-evolution of Form and Function

Evolutionary algorithms have been successfully applied to automate the design of robot morphology, the design of controllers, and more recently to coevolution of both form and function. It is our belief that the natural process of coevolving the form and function of living organisms can be applied to the design of morphology and control behaviors of autonomous vehicles in order to simplify the design process and improve the performance of the system.

In our work, coevolution of form and function has been applied to the micro air vehicle (MAV) domain. The design of the sensory payload and the controller for an MAV is complicated by the size of the vehicle (wingspan on the order of 6 inches), its limited payload, and a great variety of possible applications. The design issue currently addressed is the minimization of power requirements. It is assumed that power efficiency is inversely proportional to the sensing coverage of the sensor suite.

We have explored coevolution of the beam width and the range of the individual sensors in the sensor suite and the collision-free navigation behavior for a MAV in context of different controller representations (stimulus response rules and NN) and coevolution approaches (single- and multiple-population models). More recently, we have looked at how to adapt the components of the system (the sensor suite and the behavior) in response to unexpected changes in the environment or in their own capabilities in close to real time, which is an important problem arising for all autonomous vehicles that are expected to perform tasks for extended periods.

Multimedia

Flight of an evolved micro air vehicle system. [QuickTime] [RealPlayer]

Transitions

We have an MOU with Army Simulation, Training, and Instrumentation Command (STRICOM) to transfer research to several programs of interest to the Army, including Ground-Air Teaming Operations. This work has direct application to the autonomous operations FNC and for 6.2 D&I work within Office of Naval Research (ONR) code 351 in aircraft control. In the future, this technology could also be applied to vehicles with real-time on-line morphing capabilities.

Publications

"Anytime Coevolution of Form and Function." In Proceedings of 2003 Congress on Evolutionary Computaton (CEC-2003), Canberra, Australia, December 8 - 12, 2003.
Download PDF

"Coevolution of Form and Function in the Design of Micro Air Vehicles." In Proceedings of 2002 NASA/DoD Conference on Evolvable Hardware (EH-2002),Washington, DC, July 15-18, 2002.
Download PDF

"Co-Evolution of Form and Function in the Design of Autonomous Agents: Micro Air Vehicle Project." GECCO-2000 Workshop on Evolution of Sensors in Nature, Hardware, and Simulation, Las Vegas, NV; July 8, 2000.
Download PDF

Funding

The work reported here is supported by the ONR under work requests N0001403WR20212 and N0001403WR20057.

Contact:

Magda Bugajska, Co-Investigator

Alan C. Schultz, Co-Investigator

"Material contained herein is made available for the purpose of peer review and discussion and does not necessarily reflect the views of the Department of the Navy or the Department of Defense."