Learning Contact Dynamics for Control with Action-conditioned Face Interaction Graph Networks
This addresses the challenge of accurate contact dynamics modeling for robot control, particularly in manipulation tasks, with incremental improvements over existing methods.
The paper tackled the problem of predicting motion and force-torque for robot end effectors in contact-rich manipulation by developing a learnable physics simulator based on GNNs, achieving a 50% improvement in motion prediction accuracy and 3x increase in force-torque precision over a baseline in real-world experiments.
We present a learnable physics simulator that provides accurate motion and force-torque prediction of robot end effectors in contact-rich manipulation. The proposed model extends the state-of-the-art GNN-based simulator (FIGNet) with novel node and edge types, enabling action-conditional predictions for control and state estimation tasks. In simulation, the MPC agent using our model matches the performance of the same controller with the ground truth dynamics model in a challenging peg-in-hole task, while in the real-world experiment, our model achieves a 50% improvement in motion prediction accuracy and 3$\times$ increase in force-torque prediction precision over the baseline physics simulator. Source code and data are publicly available.