Dr. Paul Stolorz is the Technology Community Leader for Autonomy at Caltech's Jet Propulsion Laboratory, and Supervisor of the Machine Learning Systems Group, where he serves as Principal Investigator for several data mining and machine learning tasks. His work focuses on science-directed autonomy for NASA missions, scalable datamining and knowledge discovery, statistical pattern recognition, bio-computing and computational biology, with an emphasis on practical applications of novel algorithms and systems. He holds a PhD in physics from the California Institute of Technology, and a DIC from Imperial College, University of London. At JPL he has received the 1997 Lew Allen Award for Excellence, and the 1997 NASA Exceptional Achievement Medal. He serves on several program committees and editorial boards in the fields of machine learning and data mining, including service as Program Co-chair for KDD-98 (the 4th International Conference on Knowledge Discovery and Datamining), the major international meeting in this field.
A Systems Approach from Spacecraft Integration
The development of autonomous spacecraft has been a dream within several NASA engineering communities, as well as within research fields such as Artificial Intelligence, for a number of years now. In the recent past these ideas have begun to migrate towards reality as NASA's compelling need for missions with vastly lower operations costs dramatically impacts mission design. Future missions will clearly require that many fault detection and recovery functions be performed automatically onboard. Two prominent examples of this trend are the prominent success of the Sojourner rover on Mars Pathfinder, and more recently, the successful Remote Agent (RAX) and Beacon Operations (BMOX) experiments performed on the New Millenium Program mission ST1. Both missions demonstrated important advances in onboard autonomy, showing in particular important abilities to recover automatically from unexpected obstacles (Pathfinder) and unexpected instrument behavior (ST1).
In this talk I want to highlight these space mission experiences by taking up two spacecraft autonomy themes that have important implications for terrestrial autonomous systems. Firstly, I'll describe the underlying philosophy and architecture of the RAX and BMOX systems, and the lessons learned from their implementation on ST1 in terms of ensuring the health and survivability of autonomous systems in general. Secondly, I'll descibe some of the current directions of autonomy research and development at JPL and elsewhere, especially work that deals with incorporating scientific goals and data directly into autonomous systems design.
Last modified on 3/11/00 by Maggi Glasscoe (Maggi.Glasscoe@jpl.nasa.gov)