Uncertainty management and generation of geometric constraints in automated assembly systems

Shun-Feng Su, Purdue University

Abstract

This dissertation presents research regarding two problems in the automation of an assembly system: the management of uncertainties, and generation of geometric constraints for assembly tasks. For the uncertainty management, a systematic methodology of manipulating and propagating uncertainties in the task level is presented. Uncertainties are represented by covariance matrices and the manipulation of uncertainties is focused on the spatial uncertainty propagation and the uncertainty fusion. The propagations of uncertainty before and after a primitive action such as moving action, perception action, and contact action are developed. Both forward propagation and backward propagation for the verification of an action are discussed. In order to acquire the successful execution of an assembly task, goal regions are required to identify successful situations. With a normalized relational model of assemblies, goal regions are constructed by using the mating features and the constraint types residing in the relational model. Graphic simulation is performed to verify the correctness of our algorithm for constructing goal regions. An approach that shrinks the nominal boundary of a goal region is also proposed to compute the expectation of a success probability, when the boundary of the goal region is subject to uncertainties. The effects of interference and fine-motion planning on success probabilities are also taken into account to define theoretical goal regions. The result of this work is then extended to deal with the uncertainties from the assembly level point of view. The unsatisfactory situations due to the uncertainties of objects are used to identify the tolerance specification or the dimensions of the objects that are required to be modified. An algorithm with the time complexity of O(NlogN) is proposed to define the destination of a mating action which possesses the maximal estimated success probability of the mating action. This destination planner, together with the assembly trajectory planner, the grasp planner and the path planner, are the basic planners used to provide the necessary information required in assembly planning systems. (Abstract shortened with permission of author.)

Degree

Ph.D.

Advisors

Lee, Purdue University.

Subject Area

Electrical engineering

COinS