On the design of intelligent robotic agents for assembly

Robotic agents can greatly be benefited from the integration of perceptual learning in order to monitor and adapt to changing environments. To be effective in complex unstructured environments, robots have to perceive the environment and adapt accordingly. In this paper it is discussed a biology inspired approach based on the adaptive resonance theory (ART) and implemented on an KUKA KR15 industrial robot during real-world operations (e.g. assembly operations). The approach intends to embed naturally the skill learning capability during manufacturing operations (i.e., within a flexible manufacturing system). The integration of machine vision and force sensing has been useful to demonstrate the usefulness of the cognitive architecture to acquire knowledge and to effectively use it 0020-0255/$ see front matter 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ins.2004.09.011 q Thanks are sincerely due to the following organisations who have made possible this research through different funding schemes: Rolls Royce & Associates, The Nottingham Trent University, DAAD, CONACyT, CONCyTEQ and CIATEQ. * Corresponding author. Fax: +52 442 221 5243. E-mail address: [email protected] (I. Lopez-Juarez). 378 I. Lopez-Juarez et al. / Information Sciences 171 (2005) 377–402 to improve its behaviour. Practical results are presented, showing that the robot is able to recognise a given component and to carry out the assembly. Adaptability is validated by using different component geometry during assemblies and also through skill learning which is shown by the robot s dexterity. 2004 Elsevier Inc. All rights reserved.