| |||
|
|
Automated Design and Optimization for Increased Performance and Reliability on NASA Missions Is it possible to have computers automatically design and optimize complex electronic, mechanical, and nanoscale systems without being explicitly told how to do so? If so, which search algorithms are most effective? The Evolvable Systems Group investigates computer algorithms that automate the design and optimization of complex engineering systems for current and future NASA missions. Our overall goal is to dramatically increase mission reliability and science return through development and application of adaptive and evolutionary algorithms. Our current research involves writing computer algorithms that make reconfigurable electronics more reliable, design high-performance antennas and analog circuits, improve satellite scheduling, discover new robotic structures and controllers, improve multiobjective optimization via artificial coevolution, and optimize MEMS devices. Our published research is available. The nature of our work is interdisciplinary and our group members reflect this, having expertise in computer science, electrical engineering, robotics, biology, mathematics, and aerospace engineering. Our work requires vast amounts of computer power, so we've built a dedicated supercomputer for this. Some of our evolved antenna technology is headed into space on NASA's Space Technology 5 mission.
Our group organizes annual scientific
meetings on evolvable hardware technology. The next
conference
will be held in Washington DC, June 29-July 1, 2005.
|
||
