Incorporating Reactive Learning Behaviour into a Mini-robot Platform

In the Neuro-robotics Lab of the Centre for Hybrid Intelligent Systems we have organised various challenging undergraduate student projects using Mindstorm, Khepera and PeopleBot robots. In this paper we will describe a particularly interesting undergraduate student project involving the introduction of intelligent behaviour onto a mini-robot platform. In the past robots were mainly hardcoded to perform actions that limited their adaptability to their environment. As a response to this problem there has been considerable interest in producing robots that can learn. By developing a fly-catcher robot scenario using a mini-robot platform it was possible to consider the incorporation of learned intelligent behaviour for navigation and the use of object recognition. In doing so neural network learning was incorporated into a mini-robot. This work offers an insight for robot uses into issues associated with intelligent learning on mini-robots that are also applicable to those using more sophisticated robots.