Cooperative problem-solving guided by intentions and perception (abstract)

Abstract

Abstrac tThe formalization of multi-agent autonomous systems requires a rich ontolog y for capturing a variety of collective behaviours and a powerful semantics fo r distinguishing between collective agents having, executing, and jointly intendin g a plan . In this paper, we introduce the notion of social agents and social plans. A definition of joint intentions is provided that avoids some of the problem s encountered by previous formalizations . In particular, it models cooperation by requiring that agents adopt a joint goal and ajoint plan of action before formin g a joint intention . The paper also stresses the planning capability of agents an d outlines a process for means-end reasoning by multiple agents .In this paper we describe a domain independent control structure fo r cooperating problem solving. This control structure is introduced at bot h conceptual and implementational levels and is further complemented by a descriptive example . The concepts presented in this paper are derived fro m a generic agent model based upon intentions, behavior and resources o f agents. Agents are motivated to act according to their long term goals , desires, preferences, responsibilities and the like . Their actions ar e influenced by their perceived surroundings, which are other agents and th e environment in which they operate. 10

Agent-based knowledge acquisition

“…In the framework of artificial agents, in [7], the authors distinguish sensing resources, sending-receiving resources, acting resources and cognitive resources. For human agents such as experts or users, the relevant resources are rather instruments, software, tools, etc.…”

“…Of course, such low-level goals can generally be linked (and associated) to the subtasks appearing in the decomposition of the main task of the expert. In [7], the authors distinguish strategic intentions (that are long-term) and tactical intentions (that are short or mid-term). So, the knowledge engineer must elicit the possible intentions of the different experts during a problem solving, how such intentions are linked to one another, etc.…”

“…The way the individual plans of the experts can be integrated in collective plans of the organizations to which the experts belong can also be elicited, through interviews or analysis of observations, case studies or collective thinking-aloud protocols. -actions: we distinguish observable actions, actions having observable consequences, and internal actions without any observable consequence (such as the cognitive actions that are distinct from the effectoric actions, as stressed in [7]). The actions can have different levels of abstraction and different natures (e.g.…”

Specification and Implementation of a Belief-Desire-Joint-Intention Architecture for Collaborative Problem Solving

“…Burmeister and Sundermeyer have defined and implemented a generic agent model based upon intentions and desires [44]. Their agents are characterised by their perceptive capabilities, the actions they can undertake, their intentions and the resources needed for executing the different types of behaviour.…”

Meta-razonamiento en Agentes con Restricciones Temporales Críticas.

“…Otro es reactivo y se encarga de procesar los estímulos que no necesitan deliberación. De entre los agentes que poseen arquitectura híbrida podemos destacar TouringMachines [Ferguson 1992], PRS [Georgeff 1989], COSY [Burmeister 1992], AIS [Hayes-Roth 1995] e InteRRaP [Muller 1994]. …”