L-VIBRA: Learning in the VIBRA architecture. Research in Multi-Agents Systems (MAS) has been, from its outset, concerned with coordinating intelligent behavior among a collection of autonomous intelligent agents. In the last years the use of on-line learning approaches to achieve coordination has attracted an increasing attention. The purpose of this work is to use a Reinforcement Learning approach in the job of learning how to coordinate agent actions in a MAS, aiming to minimize the task execution time. To achieve this goal, a control agent with learning capabilities is introduced in an agent society. The domain on which the system is applied consists of visually guided assembly tasks such as picking up pieces, performed by a manipulator working in an assembly cell. Since RL requires a large amount of learning trials, the approach was tested in a simulated domain. From the experiments carried out we conclude that RL is a feasible approach leading to encouraging results
References in zbMATH (referenced in 3 articles , 1 standard article )
Showing results 1 to 3 of 3.
- Bianchi, Reinaldo A. C.; Costa, Anna H. R.: Comparing distributed reinforcement learning approaches to learn agent coordination (2002)
- Bianchi, Reinaldo A. C.; Costa, Anna H. R.: Ant-ViBRA: A swarm intelligence approach to learn task coordination (2002)
- Reali-Costa, Anna H.; Bianchi, Reinaldo A. C.: L-VIBRA: Learning in the VIBRA architecture (2000)