Towards Collaboration of Multiple Intelligent Robots and Human Supervision
- Dr. Inmo Jang, Postdoctoral Researcher, Robotics for Extreme Environment Group at the University of Manchester
B2 L5 R5209
As autonomy in individual robots becomes advanced, one of the next challenges is to coordinate multiple of such intelligent robots, which are then expected to innovatively transform legacy industries (e.g., warehouse automation, connected-vehicle management, etc.). Towards collaboration of multiple robots, this talk will particularly introduce a game-theoretical framework for clustering a large number of multiple robots and assigning the robot teams to given tasks, where the network of the robots is strongly connected and the individuals are asynchronous. The proposed decentralised algorithm guarantees convergence of selfish agents having social inhibition towards a Nash stable partition (i.e., social agreement) within polynomial time.
Overview
Abstract
As autonomy in individual robots becomes advanced, one of the next challenges is to coordinate multiple of such intelligent robots, which are then expected to innovatively transform legacy industries (e.g., warehouse automation, connected-vehicle management, etc.). Towards collaboration of multiple robots, this talk will particularly introduce a game-theoretical framework for clustering a large number of multiple robots and assigning the robot teams to given tasks, where the network of the robots is strongly connected and the individuals are asynchronous. The proposed decentralised algorithm guarantees convergence of selfish agents having social inhibition towards a Nash stable partition (i.e., social agreement) within polynomial time. The outcome is at least 50% suboptimal if social utilities are nondecreasing functions with respect to the number of co-working robots. The numerical experiments confirm that the proposed framework is scalable, fast adaptable against dynamical environments, and robust even in communication-disconnected situations. Furthermore, this talk will also briefly introduce a novel human-swarm interaction user interface concept that enables a single operator to supervise and guide multiple robots in a remote environment using virtual reality.
Brief Biography
Dr Inmo Jang is a postdoctoral researcher in Robotics for Extreme Environment Group at the University of Manchester. He is currently involved in RAIN(Robotics and AI in Nuclear) project, one of the four biggest robotics and AI projects funded by EPSRC, UK. His research interests range over the areas of multi-robot/agent systems, decentralised autonomous decision-making, control and navigation for robots, human multi-robot interaction, and their real-world applications. He was also a visiting researcher at the Asama Robotics Lab of the University of Tokyo, Japan. Before joining Manchester, he finished PhD in Cranfield University, UK. In South Korea, he worked at Korea Aerospace Industries, Ltd., and Korea Institute of Aviation Safety Technology in total 5+ years. Before that, he completed MSc/BSc in Mechanical and Aerospace Engineering at Seoul National University, S. Korea.