Regulation Function for Agent Adaptation Issues in Ambient Environment
- PROCEEDINGS OF THE 11TH INTERNATIONAL CONFERENCE ON COMPUTER MODELING AND SIMULATION (ICCMS 2019) AND 8TH INTERNATIONAL CONFERENCE ON INTELLIGENT COMPUTING AND APPLICATIONS (ICICA 2019)
- In this work we deal with action selection issues for software agent operating in ambient environment which is highly dynamic. Unpredictable events may occur and inappropriate actions may damage the software and its environment. Adaptation ability is a key requirement for such issues. We use multi-agent paradigm to address them. We propose a regulation function within agent architecture. It is a filter which acts on the stream of behaviour before it becomes or not action. The aim is to cope with environmental changes without the need to predict them precisely at design-time. To this end, we introduce the Influence-Reaction Model into the agent behaviour management. To facilitate its application, we implement the resulting architecture as a Java library called MECA. We experiment it with an agricultural robot moving through a field.