Abstract
This paper addresses the cyclic scheduling problem arising in two and three-machine robotic cells with time window constraints where multiple part-types are produced. Due to its complexity, very few studies have tackled the problem. Previous researches were generally limited to single or at most to two part-types where the associated part sequencing problem vanishes since the production is cyclic. We proved that the two-machine problem could be apprehended as a travelling salesman problem. For the three-machine case, we proposed a heuristic aimed at finding the sequence of robot activities and the sequence of parts that jointly minimises the cycle time in a special class of cycles. A lower bound is also provided, and computational results are reported.