Strategies for robotic handling of flexible sheet material

Abstract

Robotic manipulation strategies for handling limp flexible panels may be placed into five categories: rigidization of the panel, model-based trajectory planning, feedback control and sequentrial strategies. Sequential strategies may be either sensor-based or sensor-less. Complex tasks may best be tackled by a combination of these tactics. Rigidization techniques (e.g. jigs and adhesion) are simple, reliable and widely used but are usually limited to transferring panels between processes and are not programmable. Computational models have been developed for planning robot trajectories to drape fabric panels onto a surface without buckling the fabric. A new technique that predicts the onset of buckling when pushing a panel across a surface, permits off-line planning of sliding tasks. Robotic sewing is a dynamic task that can exhibit different buckling modes, as well as seam quality degradation, due to a dynamic and complex combination of compressive, tensile and shear forces that precludes off-line planning based on analysis. However, appropriately designed feedback control systems can provide satisfactory performance. Uncertainty in pseudo-static tasks can be reduced by a combination of sensing and sensor-less manipulation strategies. An example of a manipulation strategy for a complex handling task is given that was developed on an experimental multi-sensor robotic sewing cell.