Multi-Level Scheduling for Flexible Manufacturing1

Most real-time scheduling has focused on a relatively small set of independent tasks directly invoked periodically or via interrupts. In many real-time applications such as flexible manufacturing, this system model is too simplistic. In flexible manufacturing two entirely different sets of resources must be scheduled and “connected.” At the highest level, there are raw materials, robot arms, platform space, etc., and at the lowest level there are computational resources. Upon ordering products, high level resources must be scheduled, and the associated computational resources to achieve the manufacturing of those products must also be scheduled. This gives rise to the need for high level Real-Time Artificial Intelligence (RTAI) planners, low level schedulers that can handle large numbers of precedenceand resource-constrained tasks, and a suitable interface between the two schedulers. This paper presents the initial stages of the design and implementation of a flexible manufacturing testbed that incorporates all of these elements. 1This work is funded by the National Science Foundation under grant IRI-9208920.