A feasible collision-free and deadlock-free path-planning algorithm in a certain workspace where multiple robots move flexibly

摘要

Proposes a good collision-free and deadlock-free path-planning algorithm for multiple robots. The multiple robots act together in a same workspace simultaneously so as not to collide with each other and also not to be at a deadlock. Thus, some path-planning algorithms running in a known workspace where several robots move dynamically have been presented. However, they request much calculation time to select good collision-free and deadlock-free paths. The reason is as follows: The search space is generally huge because it is to be a three-dimensional quantization space including the time axis and also the path-planning algorithms select good paths by combinatorial investigation of their search space. For this point of view, the authors propose a good path-planning algorithm without generation of any collision or deadlock path in such a known and dynamic workspace. The algorithm can select a collision-free and deadlock-free path by a straight investigation of a nonquantization two-dimensional space not including the time axis without a fork, and consequently it is very efficient. The algorithm is designed based on asymptotical decrease of the Euclidean distance between the target robot and the goal. According to this property, the robot comes close to the goal asymptotically and consequently the robot arrives at the goal without occurrence of any deadlock.