In this paper we describe a heuristic technique for solving the 2D find-path problem for a rigid mobile body amidst a set of fixed obstacles of arbitrary shapes. The proposed approach tackles path planning as an informed search process in dis crete configuration space. Three heuristics are proposed to guide this search process, all of them relying on a global path computed from the R-MAT model of free-space (a retraction of MA T defined specifically for path planning). One of the heuristics guides the evolution of the two Cartesian degrees of freedom of the mobile body along the search, while the remaining two guide the evolution of its rotational degree of freedom. The benefits derived from the use of the proposed heuristics are twofold: on the one hand, a speed-up of the search process as compared to noninformed search algo rithms is attained, and on the other hand, the features of the resulting solution path can be somewhat controlled. Other advantages and shortcomings of the proposed path planning approach are discussed, with the practical interest of the overall process being justified by the experimental results ob tained.

No abstract

No abstract

A system allowing both the automatic programming of a robotic cell and the monitoring of assembly task execution is described. The system consists of a coordinator, a set of specialists and a task supervisor. The specialists are each associated with a particular snbtask within the assembly and have both off-line and on-line functions. The three specialists which have been incorporated into the system are grasping, displacement and insertion-The coordinator works from a task-level description of the assembly to he performed and generates a cell-centered plan based on an ordered call to the system specialists. Once the assembly operation has been planned, it is the respcmsibility of the supervisor to monitor the correct execution of the assembly task by handling multisensory integration, correcting local errors and calling--when needed--the on-line functions of the several specialists. The overall operation of the system is illustrated by means of an assembly task example. INTRODUCTIONThe application of robots in mechanical assembly tasks is a field of growing interest. Nevertheless, several open problems in the programming and execution phases of these tasks currently limit their practical industrial use. This is clear when considering that even the simplest assembly tasks require the coordinated action of different object-oriented planning systems, each of them being associated with a given elementary assembly operation. 4 Grasping,17 displacement I a and insertion t ~ constitute three examples of such elementary operations arising in most assembly tasks. Each one of these elemental planning systems has been independently studied by many research teams, 9" ~ 0. ~ 5 The results obtained so far seem to indicate the extreme complexity of this kind of apparently simple operation.