Increasing the Autonomy of Scientific Satellites to Deal With Short-Duration Phenomena

The Remote Agent Experiment (RAX) and CASPER, both from NASA, have demonstrated in flight the viability of the use of artificial intelligence (AI) planning and scheduling techniques aboard spacecrafts to increase its autonomy. Having reached full success in both cases, these systems have opened a new path for space probes and satellites applications. Following the necessity to increase the autonomy, new onboard planning systems have been being developed to deal with the difficulties of ground control and communication, the lack of resources, and the spacecraft response time for external events. RASSO, Resources Allocation Service for Scientific Opportunities, is being developed at INPE (the Brazilian National Institute for Space Research), and uses AI planning and scheduling to allow a scientific satellite to temporarily modify its current operation plan in order to better analyze short-duration phenomena. This paper describes the RASSO architecture, some of its main characteristics - as the use of the same programming language to develop the planner and describe the satellite model -, and the safe and gradual approach foreseen to validate this technology in future INPE´s satellites.