Bound to help: cooperative manipulation of objects via compliant, unactuated tails

Abstract

We examine the problem of moving multiple objects to goal locations by a coordinated team of mobile robots. Each robot is equipped with an unactuated, compliant chain attached as an appendage that we call its tail. Each of our robots tow objects by wrapping their tail around an item, securing it by hooking the end of the tail back onto itself, and then dragging. In addition to towing individually, any two robots wishing to operate within a tightly-knit sub-team are able to link the ends of their respective tails. These conjoined pairs can skim a region of space, clustering multiple objects together to transport several at once. Using operators that model both forms of towing, we formulate the planning problem for collecting multiple objects and transporting them to goal locations. We propose a general framework using logical formulas to express complex tasks. This planning problem is NP-hard and so we settle for either an exhaustive enumeration or a sub-optimal plan. The combinatorics of the action choices make the former prohibitive with as few as eight robots and objects, so we explore heuristics that give satisfactory solutions in reasonable time. We analyze the performance of the proposed algorithm to give an understanding of where it expands fewer search nodes than exact search. The results include data from physical robots executing plans produced by our planner with both individuals and coupled pairs towing objects.