Quantitative Cognitics and Agility Requirements in the Design of Cooperating Robots

Abstract

The paper makes two main theoretical contributions and also presents a cooperating robot for domestic use, as a practical study case, where the concepts agile control can be illustrated. The first contribution relates to quantitative cognitics, an approach which helps in analyzing complex situations and in designing advanced systems involving cognition, such as cooperating robots. Cognitive systems usually feature a hierarchy of subsystems. For humans, as for autonomous, cooperating robots, the functions of perception, decision, or in a more refined fashion yet, vision or trajectory planning, are good examples of (sub-) processes and tasks, which can be addressed per se, in specific subsystems. In particular, vocal communication is discussed below, as instance of a very stochastic process. For such cases, the probability of error is to be taken into account for the relevant assessment of knowledge quantities. The second contribution relates to a critical feature: relative agility of control elements. For complex cases, where many resources are necessary (e.g. groups of robots), a multi-agent structure may be useful, but then interactions occur and this involves loops, where information flows, and consequently stability becomes increasingly of concern. Several elements of solutions are proposed.