Tag
ReGuide introduces a self-improving framework for diffusion policies that uses test-time guidance to generate corrective rollouts, then fine-tunes the policy on this data, achieving 1.3–7.7× success improvement on Robomimic tasks.
Introduces TruDi, a method that enables training diffusion policies in massively parallel on-policy reinforcement learning by using a trust-region optimization rule to enforce KL constraints, achieving strong performance across 73 tasks.
QPILOTS is a method that steers flow policies at inference time by using critic gradients projected from noisy intermediate states, achieving state-of-the-art performance on offline-to-online RL benchmarks and improving pretrained VLA models without modifying the base policy.
Introduces WarmPrior, a method that replaces the standard Gaussian source in flow-matching policies with a temporally grounded prior from recent action history, consistently improving success rates on robotic manipulation tasks by producing straighter probability paths.