Search papers, labs, and topics across Lattice.
PhaForce is introduced, a phase-scheduled visual-force policy learning framework designed for contact-rich manipulation tasks. It combines a slow diffusion planner operating at low frequency with a fast corrector for high-frequency residual control, guided by a contact-aware phase predictor. The key result is an average success rate of 86% across real-robot tasks, a 40 percentage point improvement over baselines, alongside improved contact quality and robustness to geometric variations.
Achieve a 40% jump in success rates on real-world contact-rich manipulation by intelligently scheduling force feedback into visual-motor policies.
Contact-rich manipulation requires not only vision-dominant task semantics but also closed-loop reactions to force/torque (F/T) transients. Yet, generative visuomotor policies are typically constrained to low-frequency updates due to inference latency and action chunking, underutilizing F/T for control-rate feedback. Furthermore, existing force-aware methods often inject force continuously and indiscriminately, lacking an explicit mechanism to schedule when / how much / where to apply force across different task phases. We propose PhaForce, a phase-scheduled visual--force policy that coordinates low-rate chunk-level planning and high-rate residual correction via a unified contact/phase schedule. PhaForce comprises (i) a contact-aware phase predictor (CAP) that estimates contact probability and phase belief, (ii) a Slow diffusion planner that performs dual-gated visual--force fusion with orthogonal residual injection to preserve vision semantics while conditioning on force, and (iii) a Fast corrector that applies control-rate phase-routed residuals in interpretable corrective subspaces for within-chunk micro-adjustments. Across multiple real-robot contact-rich tasks, PhaForce achieves an average success rate of 86% (+40 pp over baselines), while also substantially improving contact quality by regulating interaction forces and exhibiting robust adaptability to OOD geometric shifts.
