Robustesse et Coopération en Ordonnancement – ROBOCOOP

The topic of this project is the design of cooperative approaches for robust-scheduling in a perturbed environment and their evaluation. These approaches are suitable in a lot of application fields such as production scheduling, project scheduling, grid computing, robotic activities planning, healthcare management, etc. In such environments, the execution of a schedule is highly perturbed by nature. The characteristics of activities, the temporal constraints, the resource constraints are time dependent and contingent. Therefore, building up schedules able to face a given range of uncertainties while keeping the performance under control is of prime importance with regard to the efficiency of the organization. On the other hand, because of the recent and significant progresses observed in information technologies, one can notice, in most of the previously evoked applicative fields, the development of network organization where decisions are distributed over the decision centres of the network, each one having its own decisional autonomy. This important mutation justifies the design of robust-scheduling approaches which do not consider scheduling as a centralized function any more, but rather as a cooperative function, provided that the scheduling decision arise from a clever cooperation between the different decision centres of the network. The ROBOCOOP project aims at studying, from a theoretical viewpoint, the problem of robust and cooperative scheduling, taking advantage of the various centralized robust-scheduling techniques already developed by the project partners. On the one hand, we intend to show how using robust scheduling techniques inside each decision centre can enhance the efficiency of a cooperation process, while favoring the emergence of a global organization being itself robust. On the other hand, because already existing robust-scheduling techniques have often been designed for facing a particular kind of uncertainty, we want to study how such techniques can cooperate together for providing more robust schedules. Another core topic of the ROBOCOOP project is the design and the development of evaluation tools. The aim is, apart from the evaluation of the cooperative scheduling techniques which have been previously presented, to fill an important gap that characterize robust-scheduling: the absence of evaluation tools allowing, at an international scale, the evaluation and the comparison of robust-scheduling procedures. Indeed, while some well-known scheduling problem libraries are available for comparing deterministic scheduling procedures, there does not exist any equivalent data base for robust scheduling. Yet one know how such evaluation tools favor the international competition, allow to objectively compare the efficiency of procedures and stimulate the development of more and more effective robust approaches. That is why the ROBOCOOP project aims at offering, on a totally open web site which will report the best of robust scheduling techniques, a complete evaluation chain allowing the international community to test and compare their approaches. Beyond this objective, another aim is to strengthen the position of the French community on robust scheduling and to increase its international visibility.