A distributed, operating system based, blackboard architecture for real-time control

• Authors: Daniel L. Larner
• Venue: IEA/AIE '90 Proceedings of the 3rd international conference on Industrial and engineering applications of artificial intelligence and expert systems - Volume 1
• Year: 1990

The design and implementation of a Distributed, Operating System based, Blackboard Architecture for Real-Time control (DOSBART) is described. DOSBART demonstrates the outstanding applicability of AI languages and blackboard techniques to the construction of versatile distributed real-time control frameworks. It provides a means to remotely and transparently execute operations on non-local objects, furnishing the ability to share data and blackboard structures transparently across a network of heterogeneous computers. It allows the simultaneous execution of all blackboard activities by utilizing the underlying operating system's multi-process functionality rather than its own scheduling mechanism. Architectural features are incorporated to deal with distributed real-time control issues such as interrupts, data dependencies, resource contention, activity control and I/O. Computation may occur in both process based and message based perspectives, and may be driven by a first order theorem prover that dynamically infers triggering events, by changes in demoned datums or by arbitrary predicates. The representational capabilities of the Common Lisp Object System (CLOS) were exploited to provide a rich set of base classes from which specific applications can be tailored. Lisp macro facilities support for multiple platforms.