Rotorcraft Pilot’s Associate Shared Memory Task Network Architecture

The Army’s Rotorcraft Pilot’s Associate (RPA) program demonstrates knowledge based cognitive decision aiding (CDA) technology for existing and new helicopters. The RPA system uses symmetric multiprocessing software modules (agents) connected via a shared memory context model. To coordinate agents’ modules and to provide real-time decision aiding, RPA uses a task network architecture (TNA) that includes run-time scheduling and dynamic task reallocation within a processor pool. The TNA is a software architecture or framework that supports the modeling of crew task activities, domain context, efficient prototyping, along with real-time operation in a multiprocessing system context. The TNA evolved from analysis and experience with the development and run-time requirements of the Air Force Pilot’s Associate (PA) system, subsequent Army Day/Night Advanced Pilotage System, and present RPA system. The current implementation is implemented in C++ on Silicon Graphics symmetric multiprocessing system using six of twelve processors. TNA consists of the data structures and constructs that enhance the developers’ capability to define and implement the functionality of associate systems. This paper describes the detailed design of the Task Network Architecture on the shared memory symmetric multiprocessing system.