Project planning and management using autonomous mobile agents, machine learning and negotiation protocols.

Scheduling project activities is an iterative and time-consuming process. This paper describes an innovative technique which can be used to automate these processes. The techniques described have been demonstrated through the implementation of a prototype automatic scheduler for meetings. The software uses an incremental learning algorithm to determine user preferences. The preferences are applied to a dedicated negotiation protocol to shot-ten the cycle of negotiation and to produce a result acceptable to all the users. The prototype system utilises mobile objects and autonomous agents with dedicated communication protocols. A test bed was developed to evaluate the performance of the system.