KAIST DiscoDemo：用 RL 生成多样化机器人演示，真机成功率 39%→78%
DAVIAN-Robotics · hf · 2026-10-27
KAIST and Holiday Robotics' DiscoDemo treats reinforcement learning as a data generator rather than a deployable controller to improve imitation learning:
- Idea: multiple successful solutions cover a broader success-trajectory space than repeatedly collecting one narrow solution; a privileged latent-conditioned RL generator with a parallel reverse curriculum and success-gated diversity objective discovers diverse solutions while preserving task success
- Pipeline: generated closed-loop behaviors are safety-filtered and exported as a static dataset for downstream visuomotor imitation learning
- Results: across four manipulation tasks, downstream success rose from 40% to 62% in simulation and from 39% to 78% on a real robot
「具身」频道最新
- 获宁德时代投资的 Noetix Robotics 传将赴港 IPO — pstAsiatech · 2026-10-27
- 前微软高管吐槽特斯拉 FSD:同一个路口反复犯同样的错 — MParakhin · 2026-10-27
- Meta 发布 InteractiveWorld：可验证环境测试 MLLM 能否当人类现场助手 — meta · 2026-10-27
- 近 5000 段对话转录 830 万词：MIRA 智能眼镜重度用户晒真实用量 — AnhPhuNguyen1 · 2026-10-27
- Surface Laptop Ultra 评测：本地 AI 表现亮眼但带宽受限 — draginol · 2026-10-27
- Feather 人形机器人平台售 29990 美元,主打反过度设计降本 — darian314 · 2026-10-27