Simulation-Aided Development of Multi-agent Algorithms for Tactical Missions

The development of intelligent control algorithms for robotic assets to be used in coalition operations is impeded by a significant gap between the way the theoretical artificial intelligence (A.I.) algorithms have been traditionally designed and validated and the way the practical algorithms for controlling robotic assets in (simulated) tactical missions are developed. On the first front, synthetic—usually mathematically defined— environments are used for the design and formal testing of A.I. algorithms, on the second front, low-level robotic simulators or even robotic field testing is employed. This article introduces a development process and a simulation architecture that helped us to narrow this gap. We report on how we implemented such an architecture using a software toolkit for multi-agent prototyping called Alite. The development process, the simulation architecture and the ready-made opensource toolkit Alite allowed us to develop and integrate a number of intelligent control algorithms obtained by gradual adaptation of theoretical methods to more realistic environments.