让预训练生成策略秒变力控高手，无需重训就能稳稳抓物。
UniReflex: Plug-and-Play Force Control for Pretrained Generative Policies via Fast-Slow Reflex
- 通过拦截动作头潜空间，快速生成力控响应，不碰主模型。
- 实测接触稳定性和成功率大幅提升，力控延迟降25-66倍。
- 适合需要快速力控但不想重训练的机器人应用。
生成式模仿学习策略在轨迹规划上表现优异，但缺乏闭环力控能力。直接引入力模态通常需重设计或重训练网络。本文提出UniReflex，一种通用即插即用框架，使冻结的生成策略具备可变阻抗控制（VIC）能力，用于接触调节，仅依赖演示中收集的力方向意图，无需进一步慢速骨干微调。通过非侵入式拦截动作头的深层潜在表示，UniReflex驱动一个快速反射网络，将主动施力与外部交互响应解耦，并预测归一化的各向异性刚度方向以实现定向柔顺分配。此外，该框架集成自适应门控机制，实现位置主导规划与力主导执行间的无缝切换。真实双臂实验表明，UniReflex显著提升接触稳定性与成功率，同时保持原始位置精度。在所评估的骨干模型上，其每步反向延迟比联合训练策略降低25-66倍。
原文摘要 · Abstract (English)
Generative imitation learning policies excel at trajectory planning but lack closed-loop force regulation, while directly incorporating force modalities often requires redesigning or retraining the network. We present UniReflex, a universal plug-and-play framework that equips frozen generative policies with variable impedance control (VIC) for contact regulation, guided by force-direction intent collected during demonstration, without further slow-backbone fine-tuning. By non-invasively intercepting deep latent representations from the action head, UniReflex drives a fast reflex network that decouples active force exertion from external interaction response. This scheme predicts normalized anisotropic stiffness directions for directional compliance allocation. Furthermore, UniReflex integrates an adaptive gating mechanism that enables seamless transitions between position-dominant planning and force-dominant execution. Real-world bimanual experiments demonstrate that UniReflex significantly improves contact stability and success rates while preserving original position accuracy. Our approach achieves 25-66x lower per-step backward latency relative to joint training strategies on the evaluated backbones.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。