Never Too Late for Force: Accelerating VLA Post-Training with Reactive Force Injection

预训练 VLA 具有强大的语言条件操作知识，但主要依赖视觉，在遮挡、深度模糊或微小力误差导致偏离离线分布的接触状态下容易失败。LIFT 在原动作专家旁边增加反应式动作专家，从预训练权重初始化，并通过因果力记忆和零初始化交叉注意力注入末端 6D 力，使动作能在执行中刷新。为处理策略依赖的接触分布偏移，LIFT 将反应式力注入与在线 DAgger 结合，在离线任务对齐数据和人工纠正轨迹的混合数据上训练。毛巾折叠、书本插入和河内环放置实验表明，LIFT 比仅视觉后训练学习更快、最终性能更高。

Pretrained vision-language-action (VLA) policies provide strong language-conditioned manipulation knowledge, but they remain largely vision-driven and can struggle once manipulation enters contact states where the scene is occluded, depth is ambiguous, or small force errors push execution off the offline demonstration distribution. We present LIFT, Late Reactive Injection of Force for VLA Post-Training, a force-aware post-training framework that adds contact reactivity to a pretrained VLA policy while preserving its general manipulation knowledge. LIFT grafts a reactive action expert beside the original action expert, initializes it from pretrained action weights, and injects recent 6D end-effector force through causal force memory and zero-initialized cross attention, enabling actions to be refreshed during execution. To address the policy-dependent distribution shift of contact feedback, LIFT couples reactive force injection with an online DAgger loop that trains on a mixture of offline task-alignment data and human-corrected online rollouts. Across towel folding, book insertion, and Hanoi ring placement, LIFT learns faster and reaches higher performance than vision-only post-training, while ablations show that reactive force memory and online corrective data are both important for robust contact-rich manipulation.