Deep Space 1 Ed Gamble Pandu Nayak Remote Agent Experiment Validation Remote Agent Technology Validation Objectives Test Program and Results Applicability to Future Nasa Missions Remote Agent Experiment Fact Sheet Validation Objectives

EXTENDED ABSTRACT Remote Agent is a reusable artificial intelligence (AI) software system that enables goal-based spacecraft commanding and robust fault recovery. The Remote Agent accepts high level goals from the operators and on-board software modules such as the autonomous navigator (e.g. ÒDuring the next week take pictures of the following asteroids and thrust 90% of the timeÓ). The RA determines a plan of action that will achieve those goals and carries out that plan by issuing commands to the spacecraft. The RA considers the current state of the spacecraft both in planning how to achieve its goals and in executing that plan. This allows for robust responses to failures and other contingencies. The RA detects and responds to failures in real-time and if necessary generates a new plan for achieve its remaining goals. The essential characteristics that differentiate Remote Agent from traditional flight software are that it is model-based and goal directed. In traditional software programs and expert systems, the programmer decides what the result of a program should be and writes down instructions or rules that attempt to achieve those results. The computer simply executes the instructions or fires the rules with no knowledge of what the intended result was or how it is achieving it. The term model-based refers to the fact that Remote Agent is given a model, or general description, of the behavior and structure of the spacecraft it is controlling. The term goal-directed refers to the fact that the operator tells the Remote Agent what the desired goal is. Remote Agent uses its models and reasoning algorithms to continuously find actions directed toward the goal, even if failures or anomalies occur. The Remote AgentÕs key contributions are as follows: • Goal-based commanding. The RA is commanded with high-level goals as opposed to time-ordered sequences. • Low-level commanding. Spacecraft operators can also write flexible, high level spacecraft scripts which draw upon Remote AgentÕs resource management, diagnosis, and recovery facilities to the extent desired. These allow the same level of control as sequences, but are more flexible and can draw on RA capabilities as appropriate. • Closes-loops on-board. The RA considers current conditions in deciding how to achieve the goals. • Fail-operational fault responses. Rather than always ÒsafingÓ the spacecraft, it first tries to find a way to achieve the remaining goals in spite of the failure. • The knowledge that guides the RAÕs decisions are encoded …