Several randomized path planners have been proposed dur ing the last few years. Their attractiveness stems from their applicability to virtually any type of robots, and their empir ically observed success. In this article, we attempt to present a unifying view of these planners and to theoretically explain their success. First, we introduce a general planning scheme that consists of randomly sampling the robot 's configuration space. We then describe two previously developed planners as instances of planners based on this scheme, but applying very different sampling strategies. These planners are probabilis tically complete: if a path exists, they will find one with high probability, if we let them run long enough. Next, for one of the planners, we analyze the relation between the probability of failure and the running time. Under assumptions characteriz ing the "goodness" of the robot's free space, we show that the

Several randomized p ath planners have been proposed during the last few years. Their attractiveness stems from their applicability to virtually any type o f r obots, and their empirically observed s u c cess. In this paper we attempt to present a unifying view of these planners and to theoretically explain their success. First, we introduce a general planning scheme that consists of randomly sampling the robot's con guration space. We then describe t w o p r eviously developed planners as instances of planners based on this scheme, but applying very di erent sampling strategies. These planners are p r obabilistically complete: if a path exists, they will nd one with high probability, if we let them run long enough. Next, for one of the planners, we analyze the relation between the probability of failure and the running time. Under assumptions characterizing the \goodness" of the robot's free s p ace, we show that the running time only grows as the absolute value of the logarithm of the probability of failure t h a t w e a r e willing to tolerate. We also show that it increases at a reasonable ra t e a s t h e s p ace g o odness degrades. In the last section we suggest directions for future r esearch.

Construction requires manipulating objects of various sizes and weights, including long and heavy pipes and beams. To address this need we investigate automatic multi-arm manipulation from two complementary perspectives: control and planning. This paper, which only presents one aspect of our research, focuses on the planning issues underlying multi-arm manipulation. It describes an implemented software system that computes collision-free paths for two arms in order to transport objects from their initial locations (positions and orientations) to specified goal locations. This system demonstrates several kinds of interactions between the two arms: cooperative manipulation of long objects, independent manipulation of small objects, and transfer of objects from one arm to the other to increase the workspace size.ACKNOWLEDGMENTS: This research was partially funded by CIFE (Center for Integrated Facility Engineering) and DARPA contract DAAA21-89-C0002 (Army). Y. Koga was supported in part by a Canadian NSERC Fellowship.

No abstract

In a constantly changing and partially unpredictable envi ronment, robot-motion planning must be on-line. The planner receives a continuous flow of information about occurring events and generates new plans while previously planned mo tions are being executed. This article describes an on-line planner for two cooperating arms whose task is to grab parts of various types on a conveyor belt and transfer them to their respective goals, while avoiding collisions with obstacles. Parts arrive on the belt in random order, at any time. Both goals and obstacles may be dynamically changed. This scenario is typi cal of manufacturing cells serving machine tools, assembling products, or packaging objects. The proposed approach breaks the overall planning problem into subproblems, each involving a low-dimensional configuration or configuration x time space, and orchestrates very fast primitives solving these subproblems. The resulting planner has been implemented and extensively tested in a simulated environment, as well as with a real dual- arm system. Its competitiveness has been evaluated against an oracle making (almost) the best decision at any one time; the results show that the planner compares extremely well.