Mohammadreza Chamanbaz — Research — Swarm Robotics

Swarm Robotics

One of the basic requirements of a swarm platform is to have “decentralized” communication and computation. That is, platforms should be able to establish a mesh communication network and perform all the computations locally on the platform. In a mesh network all agents are identical (from network point of view) and can sendreceive information tofrom their neighbors directly without incorporating a third agent. Therefore, loosing multiple agents in a mesh configuration does not affect the operation of others. On the other hand, in a star configuration all the information is routed using a central node. For this reason, by loosing the central node the entire network is affected and all agents loose their communication.

From computational viewpoint also it is very important not to to have a central processing node performing computations on behalf of agents. This makes agents independent form each other and the swarm will be “robust” to loss of several agents. We designed a hardware and software tools whose integration enables the technology achieving decentralized communication and computation and versatile swarm behaviors on multiple multi-robot platforms.

Please check our video where it shows a fully decentralized swarm of robots (10 units) performing a collective search and exploration. The target objective to be found is the light source (lamp in the bottom left corner) and the swarm has to deal with a completely unknown environment (with multiple obstacles).

All communications and computations are performed at the robot level in a fully decentralized fashion. This swarm of robots is highly robust to the failure of multiple units, and shows high levels of flexibility in dealing with changing environments (including dynamic ones).

This work is done in collaboration with Asst Prof. Roland Bouffanais and Asst Prof. Erik Wilhelm.

Publications

  • B. Zoss, D. Mateo, G. Tokic, M. Chamanbaz, L. Goh, F. Vallerga. R. Bouffanais, D. Yue, “Distributed System of Autonomous Buoys for Scalable Deployment and Monitoring of Large Waterbodies”, Autonomous Robots, Provisionally accepted.

  • M. Chamanbaz, D. Mateo, B. Zoss, G. Tokic, E. Wilhelm, D. Yue, R. Bouffanais, “Swarm-Enabling Technology for Multi-Robot Systems”, Frontiers in Robotics and AI, Volume 4, April 2017, Pages 1-12, ISSN 2296-9144, DOI, PDF.

  • T.G. Karimpanal, M. Chamambaz, W.Z. Li, T. Jeruzalski, A. Gupta, E. Wilhelm, “Adapting Low-Cost Platforms for Robotics Research”, in Proc. of the workshop the Path to Success: Failures in rEal Robots co-located with IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2015), Hamburg Germany, Pages 20-26, PDF.