This thesis addresses situated, embodied agents interacting in complex domains. It focuses on two problems: 1) synthesis and analysis of intelligent group behavior, and 2) learning in complex group environments. Basic behaviors, control laws that cluster constraints to achieve particular goals and have the appropriate compositional properties, are proposed as effective primitives for control and learning. The thesis describes the process of selecting such basic behaviors, formally specifying them, algorithmically implementing them, and empirically evaluating them. All of the proposed ideas are validated with a group of up to 20 mobile robots using a basic behavior set consisting of: safe--wandering, following, aggregation, dispersion, and homing. The set of basic behaviors acts as a substrate for achieving more complex high--level goals and tasks. Two behavior combination operators are introduced, and verified by combining subsets of the above basic behavior set to implement collective flocking, foraging, and docking. A methodology is introduced for automatically constructing higher--level behaviors

This paper proposes the concept of basis behaviors as ubiquitous general building blocks for synthesizing artificial group behavior in multi--agent systems, and for analyzing group behavior in nature. We demonstrate the concept through examples implemented both in simulation and on a group of physical mobile robots. The basis behavior set we propose, consisting of avoidance, safe--wandering, following, aggregation, dispersion, and homing, is constructed from behaviors commonly observed in a variety of species in nature. The proposed behaviors are manifested spatially, but have an effect on more abstract modes of interaction, including the exchange of information and cooperation. We demonstrate how basis behaviors can be combined into higher--level group behaviors commonly observed across species. The combination mechanisms we propose are useful for synthesizing a variety of new group behaviors, as well as for analyzing naturally occurring ones. Key words: group behavior, robotics, eth...

The problem of synthesizing and analyzing collective autonomous agents has only recently begun to be practically studied by the robotics community. This paper overviews the most prominent directions of research, defines key terms, and summarizes the main issues. Finally, it briefly describes our approach to controlling group behavior and its relation to the field as a whole.

This paper describes the problem of task sharing between two autonomous six--legged robots. The robots are equipped with two-way communication, object and goal sensing, and a repertoire of basic behaviors. The performance on the selected task, pushing an elongated box toward a goal region, is difficult for a single robot and improves significantly when performed cooperatively. The cooperative solution, however, requires careful coordination between the robots. We present an approach that takes advantage of a simple communication protocol to compensate for the robots' noisy and uncertain sensing and actuation, and their partial knowledge about the world. We demonstrate our approach on a series of experiments with two physical robots. 1 Introduction The work in this paper is concerned with the problem of mobile robot task sharing. In particular, we are focusing on manipulation tasks with sufficiently challenging dynamics to require the careful cooperation of two (or more) robots. In m...

Complex adaptive systems (CAS) consist of many interacting and adapting components. Echo is a computational CAS model in which evolving agents are situated in a resource-limited environment. Different views of the notion of species within Echo are compared to biological experiments on relative species abundance, specifically to Preston's "canonical" lognormal distribution.

We present a quantitative assessment of the value of cooperation for solving constraint satisfaction problems through a series of experiments, as well as a general theory of cooperative problem solving. These experiments, using both hierarchical and non-hierarchical cooperation, clearly exhibit a universal improvement in performance that results from cooperation. We also show both theoretically and experimentally the super-linear speed-up that results from having a diverse collection of skills among the cooperating agents. Our results suggest an alternative methodology to existing techniques for solving constraint satisfaction problems in computer science and distributed artificial intelligence.

Abstract Echo is a generic ecosystem model in which evolving agents are situated in a resource-limited environment. The Echo model is described, and the behavior of Echo is evaluated on two well-studied measures of ecological diversity: relative species abundance and the species-area scaling relation. In simulation experiments, these measures are used to compare the behavior of Echo with that of a neutral model, in which selection on agent genotypes is random. These simulations show that the evolutionary component of Echo makes a significant contribution to its behavior and that Echo shows good qualitative agreement with naturally occurring species abundance distributions and species-area scaling relations.

We show that when agents cooperate in a distributed search problem, they can solve it faster than any agent working in isolation. This is accomplished by having agents exchange hints within a computational ecosystem. We present a quantitative assessment of the value of cooperation for solving constraint satisfaction problems through a series of experiments. Our results suggest an alternative methodology to existing techniques for solving constraint satisfaction problems in computer science and distributed artificial intelligence.

This paper studies the properties of real-time decentralized information processing as a model of human information processing in organizations. Real-time decentralized processing---which models the computation of decision rules in a temporal decision pro blem by members of an organization---captures both the cost of computation in terms of the members' time and the constraints imposed by computational delay on the use of recent information. Unlike a batch processing model, it has no single measure of del ay because decisions are computed from data of heterogeneous lags. Furthermore, decentralization does not unambiguously reduce delay, because processing a message precludes processing current data. JEL Classifications: D83, D23 Keywords: real-time computation, decentralized information processing, organizations, bounded rationality Author's address: Department of Economics Princeton University Princeton, NJ 08544-1021 USA Voice: (609) 258-4050 Fax: (609) 258-6419 Email: tvz@Princeton....

We use several models of scale-free graphs as underlying interaction graphs for a simple model of Multi-Agent Systems (MAS), and study how fast the system reaches a fixed-point, that is, the time it takes for the system to get a 90 % of the agents in the same state. The interest of these kind of graphs is in the fact that the Internet, a very plausible environment for MAS, is a scale-free graph with high clustering and ¢ knn £ , the nearest neighbor average connectivity of nodes with connectivity k, following a power-law. Our results show that different types of scale-free graphs make the system as efficient as fully connected graphs, in a clear agreement with our previous research (Artif. Intell. 141, pp. 175-181).