CAREER: Evolvable Hardware Neuromorphic Computational Devices for Sensing and Control

Wright State University, Dayton OH

Investigators

Abstract

This is a Faculty Early Career Development (CAREER) award. The research portion of the work will develop a single, general Evolvable Hardware (EH) framework that allows for the creation of self-designing, self-configuring computational devices usable across a variety of application areas. EH seeks to create self-designing computational machines, especially those capable of exploiting design principles outside of conventionally understood engineering techniques, and self-configuring devices capable of automatically adapting as a problem domain changes. The generality of the framework will be demonstrated by using it to evolve devices to address problems drawn from several traditional computer science areas. The research plan focuses on the development of a general framework for the evolution of practical neuromorphic computational devices as well as the development of specialized VLSI chips to embody those devices. The framework will be applied to generate novel neuromorphic devices to address problems in control (jet engine combustion stabilization) and pattern recognition (active sensing for machine olfaction). By combining evolutionary learning and analog neuromorphic computing on the same chip, the project aims to deliver controllers that can adapt and improve while in service. The education plan for this project is based on the idea that professional practice requires both a strong command of theory and the skill to apply theory to solve problems of practical import. As such, engineering practice is best learned with the assistance of a mentor who can provide both theoretical knowledge and practical coaching. The educational activities are geared toward fostering mentor/student relationships among faculty, graduate students, undergraduate students, and members of the general public outside the university community. These activities will be designed to integrate research into the undergraduate and graduate curriculums, to encourage students to pursue advanced degrees in computer science and engineering, and to educate the general public in the advantages of supporting computer science and engineering research in general. This CAREER award recognizes and supports the early career-development activities of a teacher-scholar who is likely to become an academic leader of the twenty-first century. The research will have broader impacts in addition to the educational benefits. Self-designing computational machines would be of enormous value both as tools in their own right and as generators of new design principles to be exploited by humans. Self-configuring devices would be invaluable in autonomous situations where they needed to remain in service for long periods of time or in environments where human adjustment was not practical. This research aims to establish the generality of the Evolvable Hardware framework, to provide immediately useful computational devices, and to expand basic scientific knowledge.