Distributed Simulation Framework for Investigation of Autonomous Underwater Vehicles' Real-Time Behavior

The paper focuses upon framework design with distributed computing for real-time simulation of resource-intensive tasks connected with testing of autonomous underwater vehicles' (AUVs') "intelligent" control algorithms. Algorithm research is carried out by simulation of a wide range of onboard sensors, AUV dynamics, and underwater physics. A distinctive feature of the simulator is the possibility of testing algorithms directly in a selected AUV control system environment. The simulation framework architecture utilizes client-server technology and a centralized database with the possibility of connecting both personal computers and supercomputers as computational nodes. The modular approach utilized provides the possibility to expand the algorithm and functional base by connecting additional modules via plug-in technology. The framework's functional capabilities, modules' interaction diagram under the circumstances of distributed computing, data communications, and storage protocols taking into account specific features of the simulated processes are described. The framework testing results with different computational configurations illustrating the possibility of performing real-time simulations of AUV control processes with heavy computational loads are quoted.