The idea behind the helicopter was to investigate "behavior based control," and complete autonomous flight. I was very interested in subsumption architecture at the time. I did not want to repeat the work being done at MIT so I thought a helicopter would be an interesting experimental platform. Why? Because we wanted to show that we could build an intelligent, biologically inspired control system that could respond rapidly to the environment, and use a very small amount of computational resources. Of course if the robot did not respond correctly with high consistency, it would fail within second or two.

With that in mind I designed a behavior based control that was a combination of ideas from Brooks (subsumption architecture), Arbib (Schema Theory), Bekey (Reflex Control), Albus, and Arkin. The system flew and actually won some contest in Georgia a few years ago. All computation was on-board using a 68332 (a low power micro-controller).

I collaborated with Andrew Fagg [1] on the actual construction and testing of the craft as well as with many others at USC. The experiments were unbelievably hard to carry out. In the beginning, each experiment would typically result in a crash. Each crash meant extensive rebuilding. I served as the main mechanic, electronics designer, and control-architect on the design. I really began to hate the machine. The video you see is of an experiment using the gas powered craft at USC. Can you believe the smoke? DOWNLOAD QTmovie 904 k

The first design was gas powered and extremely messy to work with.

I broke off from the gas powered group and began experimenting with electric flight after visiting CGA at MIT. That visit really opened my eyes to the power of electric flight. The craft that is pictured is an all electric vehicle which is mechanically very similar to the design I saw at MIT.

Currently we have a "new and improved" version here at UCLA [2]. Here we are experimenting with the integration of visual navigation with the underlying behavioral based control. The key benefit of the electric design is that experimentation is very simple. Just flip a switch and take off. We fly indoors in the lab and have never had a crash.

I am anticipating that with the California initiative to reduce smog by forcing automobile manufacturers to build electric cars, rechargeable batteries will increase in efficiency dramatically. If we could see a 4 or 5 times increase in battery efficiency over standard NiCads we could build an all electric vehicle that could fly for 20-30 minutes.... and a miniature helicopter with a TV camera on board may be the future of news gathering in the 21st century.



[1] M. Anthony Lewis, Andrew H. Fagg, and George A. Bekey, The USC Autonomous flying Vehicle: An Experiment in Real-Time Behavior Based Control, Proceedings of the 1993 IEEE International Conference on Robotics and Automation , Atlanta, Vol II. pp 422-439, May 1993.

[2] M. A. Lewis, P. Castro, J. Hansen, S. Shell, M. J. Lamb and P. Yoo "The UCLA Autonomous Aerial Robot: System Overview," AUVS-95.