Robust Decision Making via Cooperative Estimation: Creating Data Saturated, Autonomously Generated, Simulation Environments in Near Real-Time

Abstract

Every branch of the U.S Military, as well as foreign military agents, have a vested interest in the broad applications and development of robotic systems. Advancements in data collection and storage capabilities has exposed an opportunity to increase the utility of simulated environments. At the most basic level, operations that involve robotic systems require detailed simulation environments to test algorithms and edge cases. The wealth of information collected from robotic platforms can be utilized to autonomously generate simulation environments, which can provide a robust platform for enhanced decision-making capabilities. Current industry standards depend on labor intensive post processing methods which generate static simulation environments. These simulation environments lack much utility beyond controlled testing. To address this gap, we introduce the foundational research for an intelligent simulation module, a system that utilizes sensory data, collected from semi-autonomous robotic mapping platforms, to generate in near-real time high fidelity digital twin simulation environments of real-world locations. With this system, end-users will be provided with the details they need to make operational decisions without the delay of post processing. Our system bridges the ROS platform with Unity3D game engine to achieve the generation of its simulated environments. Combat Engineer operations that rely on autonomous robotic platforms will benefit from having a system that can generate high fidelity digital twin simulation environments to aid testing research, mission planning, and robotics control. In general, the intelligent simulation system will allow for robust decision making in autonomous mobile robots, by improving navigation, path planning, coordination between agents, and task planning. This research has the potential of being utilized in hardware in the loop scenarios where multi-agent control and coordination is required to complete a mission thus advancing the field of cooperative estimation. Further, with the use of virtual reality technology, an operator could potentially be inserted into an operation site virtually; the virtual environment and agents operating within it would be parallel to the physical site, and the operator can then possibly supervise, control, and coordinate both virtual and real hardware robotic systems remotely.