Search papers, labs, and topics across Lattice.
This paper introduces FAWAM, a novel force-aware world action model that enhances robotic manipulation by integrating force information at three critical levels: perception, prediction, and closed-loop execution. By encoding historical 6-axis force/torque signals and predicting future actions alongside end-effector wrenches, FAWAM effectively models contact dynamics and refines actions in real-time through a residual correction module. Experimental results reveal a significant improvement in success rates, with a 36.25% increase over vision-only approaches and a 21.25% increase over existing force-aware methods, underscoring the model's robustness in contact-rich environments.
FAWAM achieves a 36.25% boost in success rates for robotic manipulation tasks by fully leveraging force signals to enhance action modeling and execution.
Force signals provide critical interaction cues for contact-rich robotic manipulation. However, existing methods mostly use force as an additional observation modality, without fully exploiting its role in modeling future interaction dynamics or guiding execution-time feedback correction. In this paper, we propose FAWAM, a force-aware world action model that incorporates force information at three levels: perception, prediction, and closed-loop execution. FAWAM first encodes historical 6-axis force/torque signals to modulate action generation, then jointly predicts future actions and end-effector wrenches to explicitly model contact evolution. It further introduces a residual correction module that uses the predicted wrench trajectory as an execution-time reference to refine actions online based on real-time force feedback. Real-world experiments across multiple contact-rich tasks show that FAWAM improves the average success rate by 36.25% over vision-only baselines and 21.25% over existing force-aware baselines, demonstrating the effectiveness of our force-aware framework for robust contact-rich manipulation.
