Extending the DEVS-Scheme knowledge-based simulation environment for real-time event-based control

This article describes an implementation of real-time simulation and control in DEVS-Scheme, a knowledge-based, discrete event environment. We illustrate a methodology in which the plant, its actuators and sensors are described by discrete event models developed within the event-based control paradigm. A model of the controller is employed to validate its design in a plant/actuator/sensor experimental frame. The same model con guration is then employed for actual control operation by connecting the simulation executive, suitably modi ed, to a programmable controller that interfaces to the real plant/actuator/sensor system. We show how this methodology is supported by real-time interpretation of the DEVS (Discrete Event System Speci cation) formalism. A lower bound on the processing speed of a non-deterministic operating system relative to scheduled event times is derived which guarantees correct control timing. We show how the DEVS-based control can be distributed in a hierarchical manner to ensure that the required deadline time constraints are met. As an example, an intelligent controller is discussed for a prototype oxygen extraction system eventually intended to operate autonomously on Mars. We conclude that DEVS provides a exible discrete event formalism for knowledge-based real-time control applications. 1 Research supported by NASA-Ames Co-operative Agreement No. NCC 2525, \A Simulation Environment for Laboratory Management by Robot Organization" and by The University of ArizonaNASA Space Engineering Center.