Using interaction protocols to control vision systems

In the distributed artificial intelligence (DAI) community, it is usually agreed that cooperation between reasoning agents needs explicit communication, and that the term 'communication' means more than it means in traditional distributed systems. In this paper, we propose an "interaction language" associated with "interaction protocols", inspired from the speech act theory. We demonstrate its use for the global control problem of a society of vision modules. The solution to this control problem is presented through two experimental systems, namely VAP I' and MAGIC, the basic interacting modules (or agents) of which are respectively developed at Grenoble (LIFIA) and in Nancy (CRIN).<
>