Efficient Deformable Body Simulation In Logistics: Position-Based Dynamics For Non-Rigid Packages

This study presents a novel simulation approach using Position-Based Dynamics (PBD) to model the behaviour of deformable bodies, in particular non-rigid consignments. While high-fidelity methods such as Multi-Flexible Body Dynamics (MFBD) achieve superior accuracy, their computational inefficiency is a significant barrier to large-scale applications. To address this limitation, we present a PBD-based framework that emphasizes computational efficiency without sacrificing essential physical accuracy. Using GPU-accelerated simulations via NVIDIA PhysX and IsaacSim, this study investigates the deformation and motion characteristics of non-rigid consignments using a particle-based cloth model. Experimental validation shows that PBD reduces computation time by 75% compared to MFBD for single-consignment simulations, demonstrating its efficiency. These results position PBD as a promising tool for optimizing logistics operations, particularly in automated sorting and conveyor-based handling of flexible packages. Future work will explore scalability for bulk simulations and the integration of Finite Element Method (FEM)-based models within PBD for improved accuracy.