Renewed motives for space exploration have inspired NASA to work toward the goal of establishing a virtual presence in space, through heterogeneous effets 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 constraint-based 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, NASA's rst New Millennium mission, is scheduled for a period of a week in late 1998. The development of the Remote Agent also provided the opportunity to reassess some of AI's 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.

Coalition formation is a key topic in multiagent systems. One may prefer a coalition structure that maximizes the sum of the values of the coalitions, but often the number of coalition structures is too large to allow exhaustive search for the optimal one. Furthermore, finding the optimal coalition structure is NP-complete. But then, can the coalition structure found via a partial search be guaranteed to be within a bound from optimum? We show that none of the previous coalition structure generation algorithms can establish any bound because they search fewer nodes than a threshold that we show necessary for establishing a bound. We present an algorithm that establishes a tight bound within this minimal amount of search, and show that any other algorithm would have to search strictly more. The fraction of nodes needed to be searched approaches zero as the number of agents grows. If additional time remains, our anytime algorithm searches further, and establishes a progressively lower tight bound. Surprisingly, just searching one more node drops the bound in half. As desired, our algorithm lowers the bound rapidly early on, and exhibits diminishing returns to computation. It also significantly outperforms its obvious contenders. Finally, we show how to distribute the desired

This paper analyzes coalitions among self-interested agents that need to solve combinatorial optimization problems to operate e ciently in the world. By colluding (coordinating their actions by solving a joint optimization prob-lem) the agents can sometimes save costs compared to operating individually. A model of bounded rationality is adopted where computation resources are costly. It is not worthwhile solving the problems optimally: solution quality is decision-theoretically traded o against computation cost. A normative, application- and protocol-independent theory of coalitions among bounded-rational agents is devised. The optimal coalition structure and its stability are signi cantly a ected by the agents ' algorithms ' performance pro les and the cost of computation. This relationship is rst analyzed theoretically. Then a domain classi cation including rational and bounded-rational agents is in-troduced. Experimental results are presented in vehicle routing with real data from ve dispatch centers. This problem is NP-complete and the instances are so large that|with current technology|any agent's rationality is bounded by computational complexity. 1

Anytime algorithms give intelligent systems the capability to trade deliberation time for quality of results. This capability is essential for successful operation in domains such as signal interpretation, real-time diagnosis and repair, and mobile robot control. What characterizes these domains is that it is not feasible (computationally) or desirable (economically) to compute the optimal answer. This article surveys the main control problems that arise when a system is composed of several anytime algorithms. These problems relate to optimal management of uncertainty and precision. After a brief introduction to anytime computation, I outline a wide range of existing solutions to the metalevel control problem and describe current work that is aimed at increasing the applicability of anytime computation.

Design-to-time is an approach to problem-solving in resource-constrained domains where: multiple solution methods are available for tasks, those solution methods make tradeoffs in solution quality versus time, and satisficing solutions are acceptable. Design-to-time involves designing a solution to a problem that uses all available resources to maximize the solution quality within the available time. This paper defines the design-to-time approach in detail, contrasting it to the anytime algorithm approach, and presents a heuristic algorithm for designto -time real-time scheduling. Our blackboard architecture that implements the design-to-time approach is discussed and an example problem and solution from the Distributed Vehicle Monitoring Testbed (DVMT) is described in detail. Experimental results, generated using a simulation, show the effects of various parameters on scheduler performance. Finally we discuss future research goals and plans. 1 This work was partly supported by the Of...

A Dissertation Presented by TUOMAS W. SANDHOLM

ion and Approximate Decision Theoretic Planning Richard Dearden and Craig Boutilier y Department of Computer Science University of British Columbia Vancouver, British Columbia CANADA, V6T 1Z4 email: dearden,cebly@cs.ubc.ca Abstract Markov decision processes (MDPs) have recently been proposed as useful conceptual models for understanding decision-theoretic planning. However, the utility of the associated computational methods remains open to question: most algorithms for computing optimal policies require explicit enumeration of the state space of the planning problem. We propose an abstraction technique for MDPs that allows approximately optimal solutions to be computed quickly. Abstractions are generated automatically, using an intensional representation of the planning problem (probabilistic strips rules) to determine the most relevant problem features and optimally solving a reduced problem based on these relevant features. The key features of our method are: abstractions can ...

We study task sequences that allow for speeding up the learner's average reward intake through appropriate shifts of inductive bias (changes of the learner's policy). To evaluate long-term effects of bias shifts setting the stage for later bias shifts we use the "success-story algorithm" (SSA). SSA is occasionally called at times that may depend on the policy itself. It uses backtracking to undo those bias shifts that have not been empirically observed to trigger longterm reward accelerations (measured up until the current SSA call). Bias shifts that survive SSA represent a lifelong success history. Until the next SSA call, they are considered useful and build the basis for additional bias shifts. SSA allows for plugging in a wide variety of learning algorithms. We plug in (1) a novel, adaptive extension of Levin search and (2) a method for embedding the learner's policy modification strategy within the policy itself (incremental self-improvement). Our inductive transfer case studies...

The Vickrey auction has been widely advocated for multiagent systems. First we review its limitations so as to guide practitioners in their decision of when to use that protocol. These limitations include lower revenue than alternative protocols, lying in non-private-value auctions, bidder collusion, a lying auctioneer, and undesirable revelation of sensitive information. We discuss the special characteristics of Internet auctions: third party auction servers, cryptography, and how proxy agents relate to the revelation principle and fail to promote truth-telling.

The Mars system is described which models cooperative scheduling within a society of shipping companies as a multiagent system. Emphasis is placed on the functionality of the system as a whole--- the solution of the global scheduling problem emerges from local decision-making and problem-solving strategies. An extension of the contract net protocol is presented; we show that it can be used to obtain good initial solutions for complex resource allocation problems. By introducing global information based upon auction protocols, this initial solution can be improved significantly. Experimental results are provided evaluating the performance of different cooperative scheduling strategies. Although the concepts for resource scheduling are presented solely for the transportation domain, their abstraction is useful for a broad variety of resource allocation problems. The Mars system solves the dynamic scheduling problem where no complete specification of the problem is available a priori; thus, it is designed as an on-line system based upon anytime algorithms.