Effect of low level imitation strategy on an autonomous Multi-Robot System using on-line learning for cognitive map building

Abstract

In this paper, we present the interest of low level imitation strategy on individual and population levels in the field of Multi-Robot System. Particularly, we show that adding a simple imitation capability to our bio-inspired architecture boosts the ability of individual cognitive map building. Taking into account the notion of imitative behavior, we also show that the individual discoveries in each robot (i.e. goals) could have an effect on population level and therfore it induces a new learning capability at the individual level. To analyze and validate our hypothesis, a series of experiments have been performed with and without a low level imitation strategy.

Learning by Imitation for the Improvement of the Individual and the Social Behaviors of Self-organized Autonomous Agents

“…The results show that adding an imitation capability can dramatically enhance the survival rate of the population from 45,66% (without imitation behavior) to 60,64% (with imitation). Indeed, coupling imitation behavior with the cognitive map allows agents to discover and to learn the position of the resources in the environment more rapidly We have also validated the positive feedback of the imitation strategy on a minimal robotics setup [13] based on the same bio-inspired architecture which has been validated in real robots in [14] . We conducted 20 experiments (this number is determined by the statistical Fisher test) composed of an imitator robot IR (having an imitation behavior) and with a leader robot LR (that has already learned the position of the goals G1 and G2 with a threshold of vigilance to learn new places equal to 0.65 and a duration of learning equal to 30 minutes).…”

Optimization through coupling learning capability and imitation strategy in a Multi-Robot System

“…Therefore modern robotics considers imitation as a powerful behavior that enables learning by observation [4], [5]. A series of experiments has been performed to validate that the imitation strategy improves multi-robot system [6], [7], [8], [9], [10], [11].…”