A Framework based on Real-Time OS and Multi-Agents for intelligent autonomous robot competitions

TitleA Framework based on Real-Time OS and Multi-Agents for intelligent autonomous robot competitions
Publication TypeConference Paper
Year of Publication2016
AuthorsCalvaresi, D., P. Sernani, A. Claudi, A. Balsini, M. Marinoni, A F. Dragoni, and G. Buttazzo
Conference Name11th IEEE International Symposium on Industrial Embedded Systems
KeywordsMulti-Agent Systems, Path-planning, Real-time control, Robot Competition

Robots interacting with human beings are widespread in modern environments, and those performing intelligent tasks without human supervision need to take into account potential criticalities. Making robots compete enable their evaluation with respect to navigation, mapping, object recognition, tracking and manipulation capabilities. Robot competitions date back to the early '80s proving to be useful for educational and research purposes. Several competitions are focused on human-robot interaction, even though they rarely produce as outcome robots capable to seamlessly interact with human beings. The main reason for this is the lack of understanding of human intentions and the failure to rapidly react to human actions. In other words, an ideal robot must be able to communicate and coordinate with humans or with other robots, to act autonomously, and to react under real-time constraints. This paper proposes a new framework to simplify the development of intelligent robots, testing them in a real robot competition. The framework combines (i) a multi-agent system to interact with humans, other robots and perform object identification and pathfinding, and (ii) a real-time motion control deployed on the Erika RTOS, to move the robot and react in a timely fashion to changes in the environment. In the considered competition scenario, the robot is required to identify and collect common objects in a bounded arena with dynamic obstacles in a limited amount of time, receiving commands from humans and competing with other robots. This approach confirms the powerful combination of multi-agent systems, computer vision, and real-time systems.