Simulation and Optimization of Industrial Disassembly Paths using Position Based Dynamics

Abstract

The finding of plausible assembly paths of industrial components is a relevant, actual, but not yet fully resolved research topic. For a fast and robust computation of such paths, rigid body sampling-based motion planning is used. To tackle the problem of unavoidable collisions due to overpressure of components or due to flexible fastening elements, the tolerance of minor collisions is necessary. Tolerating minor collisions always means the manipulation of the configuration space or the direct manipulation of objects in the workspace. These actions enable the computation of assembly paths on industrial data, but they affect (explicitly or implicitly) the shape of the objects and thus the physical plausibility of the computed paths. This makes it important to further analyze the assembly paths. In this paper, we propose a postprocessing method for indus-trial assembly paths that is based on Position Based Dynamics (PBD) and is able to simulate and optimize given assembly paths. We use the PBD framework to simulate and measure the object deformation along the assembly path. For the optimization of the path, we compare the deformed object to the original rigid object and we apply small corrections to the path that decrease the overall deformation. We show the effectiveness of our approach on an academic dataset that provides industrial disassembly scenarios.