In a distributed algorithm, multiple processes, or agents, work toward a common goal. More often than not, the actions of some agents are dependent on the previous execution (if not also on the outcome) of the actions of other agents. The resulting interdependencies between the timings of the actions of the various agents give rise to the study of methods for timely coordination of these action...

Conservative design is the ability of an individual agent to ensure predictability of its overall performance even if some of its actions and interactions may be inherently less predictable or even completely unpredictable. In this paper, we describe the importance of conservative design in cooperative multi-agent systems and briefly characterize the challenges that need to be addressed to achi...

Coordinating multiple agents that need to perform a sequence of actions to maximize a system level reward requires solving two distinct credit assignment problems. First, credit must be assigned for an action taken at time step t that results in a reward at time step t′ > t. Second, credit must be assigned for the contribution of agent i to the overall system performance. The first credit assig...

We consider the problem of coordinating the behavior of multiple self-interested agents. It involves constraint optimization problems that often can only be solved by local search algorithms. Using local search poses problems of incentivecompatibility and individual rationality. We thus define a weaker notion of bounded-rational incentive-compatibility where manipulation is made impossible with...

This paper is intended to give an intuitive overview of the operations of MASA-CIRCA, the Multi-Agent SelfAdaptive Cooperative Intelligent Real-Time Control Architecture. While individual CIRCA agents have been under development for some time, we have only recently begun developing the architecture’s multiagent capabilities. This paper briefly describes the high-level negotiation functions that...

Agents' behaviour is not only dictated by utility calculus, as has frequently been proposed. Even if we adopt a welfarist design, interests and not utilities seem to guide a very important part of agents' behaviour in Multi-Agent Systems (MAS). Agents have rights that protect their interests against additive utilities: Rights allow them enough freedom, but at the same time constrain them (prohi...

Many computer systems are comprised of multiple entities (or agents) with common objectives. Though these systems can be made intelligent, using artificial intelligence techniques, individual agents are often restricted in their capabilities and have only limited knowledge of their environment. However, the group as a whole is capable of executing more complex tasks than a single agent can perf...

Transfer learning leverages an agent’s experience in a source task in order to improve its performance in a related target task. Recently, this technique has received attention in reinforcement learning settings. Training a reinforcement learning agent on a suitable source task allows the agent to reuse this experience to significantly improve performance on more complex target problems. Curren...

This paper describes and discusses a novel design and a prototype implementation for manufacturing control systems, aimed at handling changes and disturbances. This novel design utilizes the concept of a multi-agent system. Agents in this system use an indirect coordination mechanism, called stigmergy. Stigmergy is a class of mechanisms that mediate animal-animal interactions. It consists of in...

Many challenges in multi-agent coordination can be modeled as distributed constraint optimization problems (DCOPs) but complete algorithms do not scale well nor respond effectively to dynamic or anytime environments. We introduce a transformation of DCOPs into graphical games that allows us to devise and analyze algorithms based on local utility and prove the monotonicity property of a class of...