Subdimensional expansion for multirobot path planning

Abstract

Planning optimal paths for large numbers of robots is computationally expensive. In this paper, we introduce a new framework for multirobot path planning called subdimensional expansion, which initially plans for each robot individually, and then coordinates motion among the robots as needed. More specifically, subdimensional expansion initially creates a one-dimensional search space embedded in the joint configuration space of the multirobot system. When the search space is found to be blocked during planning by a robotrobot collision, the dimensionality of the search space is locally increased to ensure that an alternative path can be found. As a result, robots are only coordinated when necessary, which reduces the computational cost of finding a path. We present the M* algorithm, an implementation of subdimensional expansion that adapts the A* planner to perform efficient multirobot planning. M* is proven to be complete and to find minimal cost paths. Simulation results are presented that show that M* outperforms existing optimal multirobot path planning algorithms.

Spatio-temporal decomposition: a knowledge-based initialization strategy for parallel parking motion optimization

“…This pioneering work inspired subsequent studies to plan in low-dimensional search spaces initially [31,32]. You et al [33] introduced circumcircles to cover rectangular vehicles, aiming to ease the collision-avoidance descriptions.…”

Effective Metrics for Multi-Robot Motion-Planning

“…This is why the study continues henceforth with dRRT [44]an adaptation of RRT to the multi-robot setting, which can cope with a larger number of robots and more complicated tasks than RRT as-is. We mention that M* [51], which is another sampling-based planner tailored for MRMP, is less relevant to our current discussion since it only employs metrics concerning individual robots.…”

“…This approach usually comes with stronger theoretical guarantees such as completeness [29,39,40,41,44] or even optimality [51] of the returned solutions. However, due to the computational hardness of MRMP [21,22,26,42,45], coupled techniques do not scale well with the increase in the number of robots.…”

“…Sampling-based planners have proven to be effective in challenging settings of the single-robot case, and a number of such planners have been proposed for MRMP [44,50,51]. Sampling-based planners attempt to capture the connectivity of the free space by sampling random configurations and connecting nearby configurations by simple collision-free paths.…”

AI buzzwords explained

“…The tighter the space, the longer the runtime. Researchers have also suggested a variety of other multi-agent path-finding techniques (Silver, 2005;Sturtevant & Buro, 2006;Ryan, 2008;Wang & Botea, 2008Luna & Bekris, 2011;Sharon, Stern, Goldenberg, & Felner, 2013;de Wilde, ter Mors, & Witteveen, 2013;Barer, Sharon, Stern, & Felner, 2014;Goldenberg et al, 2014;Wagner & Choset, 2015;Boyarski et al, 2015;Cohen et al, 2016), including some that transform the problem into a different problem for which good solvers exist, such as satisfiability (Surynek, 2015), integer linear programming (Yu & LaValle, 2013b) and answer set programming (Erdem, Kisa, Oztok, & Schueller, 2013). Researchers have also studied how to execute the resulting solutions on actual robots (Cirillo, Pecora, Andreasson, Uras, & Koenig, 2014;Hoenig et al, 2016 (Wang, 2012;Sharon, 2016).…”