Using the DIMMACSS-PSG Intelligent Robotic Middleware to Control Real-World and Simulated Multi-Agent Systems

摘要

DIMMACSS-PSG is a free and open-source intelligent robotic middleware that allows a Simulink controller to control a physical real-world multi-agent system or a simulated multi-agent system existing within the Player/Stage/Gazebo general-purpose robotic simulator, using the exact same control system for both in an optimal and efficient manner. A general-purpose robotic simulator, such as Player/Stage/Gazebo, is designed to simulate everything inherent in a real-world robotics experiment, including all types of robotic hardware such as actuators and sensors (stereo-vision camera's, moving/rotating parts, motors, etc), and provides realistic environments, physics, and sensor data. Intelligent robotic middleware, such as DIMMACSS-PSG, is the class of technologies that sit between a theoretical algorithm/function and the target real-world or simulated devices, and is required in order to actually realize an application such as one implemented on an unmanned aerial vehicle or a surface exploration robot. This paper discusses the motivation for DIMMACSS-PSG, details the important design decisions, and presents a demonstration of a Simulink controller that uses DIMMACSS-PSG to control both a simulated multi-agent system and a real-world multi-agent system.