Search papers, labs, and topics across Lattice.
This paper introduces PhysCoRe, a physics-corrected residual world model that integrates a differentiable Material Point Method (MPM) simulator with two neural networks to enhance the prediction of deformable object dynamics during robotic manipulation. By employing a Material from Motion (MfM) module to infer elasticity from visual data and a Residual from Dynamics (RfD) module to correct systematic biases, PhysCoRe achieves superior prediction accuracy compared to existing methods. The model not only adapts to new materials with limited data but also provides a confidence distribution that guides further exploration, marking a significant advancement in material-aware robotics.
PhysCoRe outperforms traditional methods by accurately predicting deformable object dynamics while adapting to new materials and guiding exploration through confidence-based feedback.
Predicting how deformable objects evolve under robotic manipulation is a longstanding challenge. Existing approaches typically rely on per-object optimization to fit material parameters, which can be slow and cannot generalize, while end-to-end learned alternatives extrapolate poorly and often violate basic physical structure. We present PhysCoRe, a physics-corrected residual world model that couples a differentiable Material Point Method (MPM) simulator with two feed-forward neural networks. A material refinement module, Material from Motion (MfM), infers per-particle elasticity from visual observations, grounding the simulator in object-specific physics. A residual correction module, Residual from Dynamics (RfD), learns the discrepancy and predicts corrections to the simulator's internal dynamics, absorbing systematic biases that the analytical model cannot capture. This design also supports online material identification on novel objects. MfM adapts from limited interactions, and its predictive uncertainty steers further exploration toward the regions where its estimate is least confident. Experiments on real deformable-object manipulation sequences show that PhysCoRe outperforms state-of-the-art baselines in prediction accuracy, and that its predicted confidence forms a reliable distribution across the object's geometry, providing a natural signal for future confidence-guided exploration.
