Congratulations to Andrew Dobson, Rahul Shome and Prof. Kostas Bekris as well as their Tel Aviv University collaborators, Kiril Solovey and Prof. Dan Halperin, for receiving the Best Paper Award at the 1st IEEE-RAS International Symposium on Multi-Robot and Multi-Agent Systems (MRS), which took place Dec. 4-5, 2017 on the University of Southern California campus in Los Angeles, CA. The acceptance rate of the conference was 24%.
A video showing two dual-arm manipulators moving in the same environment according to the output of the proposed approach dRRT* in the paper is available here: https://www.youtube.com/watch?v=ePs03FKCBrE&feature=youtu.be
The paper shows that dRRT* converges to asymptotically optimal paths when planning for the simultaneous motion of multiple robots in a common workspace. This is achieved despite the fact that during a preprocessing phase data structures for storing trajectories of the robots, i.e., probabilistic roadmaps, are built separately for each individual robot. During the online phase, the method searches implicitly the tensor product of the individual robot probabilistic roadmaps in order to discover paths of high-quality that do not lead into collisions between the robots. The method is shown experimentally to scale to challenges with many degrees of freedom where competitive state-of-the-art methods that can provide similar guarantees do not scale.
This paper is a collaborative effort with Prof. Dan Halperin's group from Tel Aviv University and is extending prior work in multi-robot planning by the Tel Aviv University co-authors Kiril Solovey (PhD student) and Dan Halperin. Since the submission, the first author of the paper, Andrew Dobson, has graduated with a PhD degree from the Computer Science department at Rutgers and has joined the Univ. of Michigan, Ann Arbor as a postdoctoral researcher. He gave the presentation during the conference in Los Angeles. Rahul Shome is continuing his PhD studies in the Computer Science department.