This special issue of Recherche Operationnelle is devoted to recent research which significantly enhances our ability to model, either mathematically or algo-rithmically, increasingly complex service systems. All the papers in this issue are methodological in nature, even though the first paper addresses a very practical approach. This collection of papers begins with a survey by Dr Khaled Hussain and Mr Varol Kaptan of recent advances which allow modelling and simulation to be conducted in a very realistic visual setting. We all know how frustrating it can be to deal with the need to determine whether the objects we model are represented in a realistic manner, and whether the simulation conditions are indeed realistic. The approach described in this first paper is based on the premise that the system being simulated has a realistic visual counterpart, e.g. some natural scenery (as in a military tactical simulation), a room (as when people's behaviour may be simulated), some roads or an airport (as when a transportation system has to be simulated), a hospital operating room (as when a surgical procedure needs to be simulated), and so on. In all of these cases, the real world being simulated can be based on an actual video sequence of the real world setting. It can be even more realistic and the simulation can be conducted within a video viewing of the real world itself. This paper shows how this can actually be done and describes a system that has been designed and developed at the University of Central Florida and at Imperial College to achieve this. In such a simulator, the simulated conditions are the real world itself, while the behaviour of synthetic objects or synthetic changes to real objects is what is being simulated. The value of such simulations can be tremendous, and can allow testing of complex behaviours in a fully realistic setting with the possibility of the human viewer actually observing different operating conditions. In addition to the visual aspects, the authors also discuss how different objects in the simulation can be coordinated using on-line learning and control algorithms that imitate intelligent reactive behaviours, based on both deterministic and randomized algorithms.
展开▼
