Recently the interest in theoretical aspects of planning, such as soundness, completeness and complexity of planners, has grown. The research related to automated planning and scheduling in Artificial Intelligence has followed various directions: tele-communication, management of temporal constraint networks, and timetabling are the most interesting problem of planning investigated so far.
The development of a specialized module to deal with information about time in a planning or scheduling architecture has been considered. Particular attention has been given to the representation of quantitative information and to the design of dynamic algorithms that allow to incrementally post and remove constraint from the network. A specialization of the AC-3 algorithm for constraint satisfaction has been proposed that allows effective on-line management of temporal constraints. Present results are described in [CCO94].
The DRS-Sched system is a knowledge-based scheduling system for planning the activities of a particular satellite system (the DRS). The system represents a peculiar set of constraints given by the domain problem and uses specialized heuristics to produce effective schedules. Different aspects of the system are described in [AC94b][AC94a][AC95] These activities are supported by CNR Special Project on Automated Planning.
Communication between agents and the planning of load balancing in a multi agent environment have also been tackled in [SYT95] with a particular focus on learning. The problem is addressed for systems which can relay neither on central coordination nor on explicit communication. Results have been found on the interplay between basic adaptive behavior parameters and the system efficiency have been produced. The properties of adaptive load balancing in heterogeneous populations are also investigated.
Timetabling is another interesting problem studied in the last years. It can be described in terms of a fixed a sequence of meetings between teachers and students in a prefixed period of time (typically a week), satisfying a set of constraints of various types. This problem, that has been traditionally considered in the operational research field, has recently been tackled with techniques belonging also to artificial intelligence (e.g. genetic algorithms, tabu search, and simulated annealing). The paper [Sch95a] is a survey of various formulations, techniques and algorithms used for its solution.