Barney Pell

Remote Agent: to boldly go where no AI system has gone before

Abstract Renewed motives for space exploration have inspired NASA to work toward the goal of establishing a virtual presence in space, through heterogeneous fleets of robotic explorers. Information technology, and Artificial Intelligence in particular, will play a central role in this endeavor by endowing these explorers with a form of computational intelligence that we call remote agents . In this paper we describe the Remote Agent, a specific autonomous agent architecture based on the principles of model-based programming, on-board deduction and search, and goal-directed closed-loop commanding, that takes a significant step toward enabling this future. This architecture addresses the unique characteristics of the spacecraft domain that require highly reliable autonomous operations over long periods of time with tight deadlines, resource constraints, and concurrent activity among tightly coupled subsystems. The Remote Agent integrates constraintbased temporal planning and scheduling, robust multi-threaded execution, and model-based mode identification and reconfiguration. The demonstration of the integrated system as an on-board controller for Deep Space One, NASAs first New Millennium mission, is scheduled for a period of a week in mid 1999. The development of the Remote Agent also provided the opportunity to reassess some of AIs conventional wisdom about the challenges of implementing embedded systems, tractable reasoning, and knowledge representation. We discuss these issues, and our often contrary experiences, throughout the paper.

The Belief-Desire-Intention Model of Agency

Within the ATAL community, the belief-desire-intention (BDI) model has come to be possibly the best known and best studied model of practical reasoning agents. There are several reasons for its success, but perhaps the most compelling are that the BDI model combines a respectable philosophical model of human practical reasoning, (originally developed by Michael Bratman [1]), a number of implementations (in the IRMA architecture [2] and the various PRS-like systems currently available [7]), several successful applications (including the now-famous fault diagnosis system for the space shuttle, as well as factory process control systems and business process management [8]), and finally, an elegant abstract logical semantics, which have been taken up and elaborated upon widely within the agent research community [14, 16].

An autonomous spacecraft agent prototype

This paper describes the New Millennium Remote Agent (NMRA) architecture for autonomous spacecraft control systems. The architecture supports challenging requirements of the autonomous spacecraft domain not usually addressed in mobile robot architectures, including highly reliable autonomous operations over extended time periods in the presence of tight resource constraints, hard deadlines, limited observability, and concurrent activity. A hybrid architecture, NMRA integrates traditional real-time monitoring and control with heterogeneous components for constraint-based planning and scheduling, robust multi-threaded execution, and model-based diagnosis and reconfiguration. Novel features of this integrated architecture include support for robust closed-loop generation and execution of concurrent temporal plans and a hybrid procedural/deductive executive.

Issues in temporal reasoning for autonomous control systems

Deep Space One will be the rst spacecraft to be controlled by an autonomous agent poten tially capable of carrying out a complete mission with minimal commanding from Earth The New Millennium Remote Agent NMRA includes a planner scheduler that produces plans and an executive that carries them out In this paper we discuss several issues arising at the interface between planning and execution including exe cution latency plan dispatchability and the dis tinction between controllable and uncontrollable events Temporal information in the plan is rep resented within the general framework of Simple Temporal Constraint networks as introduced by Dechter Meiri and Pearl However the execu tion requirements have a substantial impact on the topology of the links and the propagation through the network

A strategic metagame player for general chess-like games

This paper introduces METAGAMER, the first program designed within the paradigm of Metagame‐playing (Metagame). This program plays games in the class of symmetric chess‐like games, which includes chess, Chinese chess, checkers, draughts, and Shogi. METAGAMER takes as input the rules of a specific game and analyzes those rules to construct an efficient representation and an evaluation function for that game; they are used by a generic search engine. The strategic analysis performed by METAGAMER relates a set of general knowledge sources to the details of the particular game. Among other properties, this analysis determines the relative value of the different pieces in a given game. Although METAGAMER does not learn from experience, the values resulting from its analysis are qualitatively similar to values used by experts on known games and are sufficient to produce competitive performance the first time METAGAMER plays a new game. Besides being the first Metagame‐playing program, this is the first program to have derived useful piece values directly from analysis of the rules of different games. This paper describes the knowledge implemented in METAGAMER, illustrates the piece values METAGAMER derives for chess and checkers, and discusses experiments with METAGAMER on both existing and newly generated games.

A hybrid procedural/deductive executive for autonomous spacecraft

The New Millennium Remote Agent (NMRA) will be the first AI system to control an actual spacecraft. The spacecraft domain places a strong premium on autonomy and requires dynamic recoveries and robust concurrent execution, all in the presence of tight real-time deadlines, changing goals, scarce resource constraints, and a wide variety of possible failures. To achieve this level of execution robustness, we have integrated a procedural executive based on generic procedures with a deductive model-based executive. A procedural executive provides sophisticated control constructs such as loops, parallel activity, locks, and synchronization which are used for robust schedule execution, hierarchical task decomposition, and routine configuration management. A deductive executive provides algorithms for sophisticated state inference and optimal failure recovery planning. The integrated executive enables designers to code knowledge via a combination of procedures and declarative models, yielding a rich modeling capability suitable to the challenges of real spacecraft control. The interface between the two executives ensures both that recovery sequences are smoothly merged into high-level schedule execution and that a high degree of reactivity is retained to effectively handle additional failures during recovery.

Mission operations with an autonomous agent

The Remote Agent (RA) is an Artificial Intelligence (AI) system which automates some of the tasks normally reserved for human mission operators and performs these tasks autonomously on-board the spacecraft. These tasks include activity generation, sequencing, spacecraft analysis, and failure recovery. The RA will be demonstrated as a flight experiment on Deep Space One (DS1), the first deep space mission of the NASAs New Millennium Program (NMP). As we moved from prototyping into actual flight code development and teamed with ground operators, we made several major extensions to the RA architecture to address the broader operational context in which RA would be used. These extensions support ground operators and the RA sharing a long-range mission profile with facilities for asynchronous ground updates; support ground operators monitoring and commanding the spacecraft at multiple levels of detail simultaneously; and enable ground operators to provide additional knowledge to the RA, such as parameter updates, model updates, and diagnostic information, without interfering with the activities of the RA or leaving the system in an inconsistent state. The resulting architecture supports incremental autonomy, in which a basic agent can be delivered early and then used in an increasingly autonomous manner over the lifetime of the mission. It also supports variable autonomy, as it enables ground operators to benefit from autonomy when they want it, but does not inhibit them from obtaining a detailed understanding and exercising tighter control when necessary. These issues are critical to the successful development and operation of autonomous spacecraft.

Abstract resource management in an unconstrained plan execution system

We describe the abstract resource management mechanism in ESL (execution support language). ESL is the implementation substrate for the New Millennium Remote Agent Smart Executive, part of a NASA program to demonstrate autonomous control of an unmanned spacecraft scheduled to launch in 1998. The executive is responsible for robust plan execution in the face of unexpected run-time contingencies. Part of this task requires run-time management of the spacecrafts configuration, whose component states are modeled as abstract resources. In this paper we describe the ESL constructs for managing these abstract resources. The resource management facilities in ESL are similar to the constraint management constructs in RAPs. The major contribution in this paper is the implementation of these facilities in an unconstrained execution substrate implemented as an extension to a standard programming language (in this case, Common Lisp) rather than within a constrained self-contained plan execution language. This turns out to significantly simplify complex programming tasks. The main technical problem in a resource management system is designing a representation that allows automatic determination of when conflicts exist. In its full generality this becomes a full-blown planning problem, and therefore an impractical strategy for a reactive executive. Instead, we model abstract resources as properties, logical assertions whose final values are guaranteed unique. When two properties are identical but for their final value then they are in conflict. This paper describes the ESL constructs and mechanisms for scheduling tasks so that they do not attempt to achieve conflicting properties simultaneously, and for invoking external recovery mechanisms for restoring properties to their desired states when forced away from those states by unexpected contingencies.

Infusion of autonomy technology into space missions: DS1 lessons learned

The impact of infusing breakthrough autonomy technology into a flight project was a big surprise. Valuable technical and cultural lessons, many of general applicability when introducing system-level autonomy, have been learned by infusing the Remote Agent (RA) into NASAs Deep Space 1 (DS1) spacecraft. The RAs architecture embodies system-level autonomy in three major components: planning and scheduling, execution, and fault diagnosis and reconfiguration. Lessons learned include: the architecture was confirmed; active participation by nonautonomy personnel in the development is essential; communication of new concepts is essential, difficult, and hampered by differences in terminology; giving a spacecraft system-level autonomy changes organizational roles in operating the spacecraft after launch, and hence changes roles during development; software models supporting functions traditionally handled on the ground must be developed early enough to get on-board; shortfalls in planned features must be technically and developmentally accomodatable, in particular not to threaten the launch schedule; traditional commanding must be supported; testing must be emphasized. These lessons and others, on incremental system releases and use of autocode generation, are based on 16 months of spiral development from start of project through the projects decision to reduce the role of the RA from full-time control of the spacecraft to a separable experiment.

Smart executives for autonomous spacecraft

To make spacecraft more self-reliant and less dependent on ground intervention, the authors have developed an executive (EXEC) for a new control architecture, Remote Agent. EXEC has an expanded vocabulary for commanding spacecraft, provides a wide range of capabilities at different levels of abstraction, and has a highly modular design.