Robust Autonomy
Papers from the AAAI Spring Symposium
Ella M. Atkins and Lorraine M. Fesq,Cochairs
Technical Report SS-01-06
134 pp., $30.00
ISBN 978-1-57735-145-0
[Add to Cart] [View Cart]
AI researchers have made exciting progress in designing architectures that perform basic functions required for intelligent autonomous operations. With the potential for lowering operations costs and enabling new capabilities, there is a growing interest in integrating these architectures into our everyday lives. However, when these architectures are deployed, operators often discover that the resulting systems don't quite work as planned when challenging situations are encountered. In fact, a general perception is that increased autonomy equates to increased risk. As automation extends into ambitious domains such as spacecraft and reactor process control, the cost of not working quite as planned is enormous. In these domains, it is important that any controller be reliable and robust to the spectrum of possible situations, even those involving systems failure or unexpected events. This symposium explored the issues and approaches involved in applying AI techniques to domains in which robust and reliable operation is critical.