LoopWAM World Action Model Release and RoboChallenge Table30 V2 Top Performance
Overview
This paper presents LoopWAM, a closed-loop world-action model that learns directly from full-quality deployment rollouts, including successes, suboptimal behavior, and failures. By integrating source-dependent noise sampling, quality steering, and quality-weighted supervision, LoopWAM achieves significant improvements in success rates on benchmarks like RoboChallenge Table30-V2, RoboTwin 2.0, and RoboCasa365, demonstrating effective closed-loop policy improvement under both multi-task learning and single-task post-training scenarios.
From arXiv Robotics — research abstracts
Event evidence and corrections
1 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
Reported quantity · Reported success rate: 60 percent · Basis not reported · Differing source assertions
“在本次评测中，LoopWAM在11项任务上的成功率达到 60% 及以上，9项任务过程分达到80分及以上。”
Exact source · revision 1jiqizhixin
Reported quantity · Reported success rate: 100 percent · Basis not reported · Differing source assertions
“几组具体成绩，让模型能力更加直观： 整理水果、物品分类、取出绿色积木：成功率均为 100% 。”
Exact source · revision 1jiqizhixin
Reported quantity · Reported success rate: 90 percent · Basis not reported · Differing source assertions
“用小勺舀取物体：成功率 90% ，过程分 92 。”
Exact source · revision 1jiqizhixin
Reported quantity · Reported success rate: 80 percent · Basis not reported · Differing source assertions
“按下按钮：成功率 80% ，过程分 92.5 。”
Exact source · revision 1jiqizhixin
Reported quantity · Reported success rate: 80 percent · Basis not reported · Differing source assertions
“堆叠碗具：成功率 80% ，过程分 89 。”
Exact source · revision 1jiqizhixin
Reported quantity · process score: 92 other · Basis not reported · Differing source assertions
“用小勺舀取物体：成功率 90% ，过程分 92 。”
Exact source · revision 1jiqizhixin
Reported quantity · process score: 92.5 other · Basis not reported · Differing source assertions
“按下按钮：成功率 80% ，过程分 92.5 。”
Exact source · revision 1jiqizhixin
Reported quantity · process score: 89 other · Basis not reported · Differing source assertions
“堆叠碗具：成功率 80% ，过程分 89 。”
Exact source · revision 1jiqizhixin
Artifact availability · code: available · Differing source assertions
“项目主页： https://loop-wam.github.io 论文原文： https://arxiv.org/abs/2610.18873 开源代码： https://github.com/loop-wam/LoopWAM 近日，由 清华大学、原力灵机与 极寰时代 联合研发的世界动作模型LoopWAM，登顶真机评测榜单RoboChallenge Table30 V2，成功率与过程分双双位列榜首，成绩超过VLA-DM0.5等同期方案。”
Exact source · revision 1jiqizhixin
Artifact availability · paper: available
“项目主页： https://loop-wam.github.io 论文原文： https://arxiv.org/abs/2610.18873 开源代码： https://github.com/loop-wam/LoopWAM 近日，由 清华大学、原力灵机与 极寰时代 联合研发的世界动作模型LoopWAM，登顶真机评测榜单RoboChallenge Table30 V2，成功率与过程分双双位列榜首，成绩超过VLA-DM0.5等同期方案。”
Exact source · revision 1jiqizhixin
Artifact availability · code: available · Differing source assertions
“项目主页： https://loop-wam.github.io 论文原文： https://arxiv.org/abs/2610.18873 开源代码： https://github.com/loop-wam/LoopWAM 近日，由 清华大学、原力灵机与 极寰时代 联合研发的世界动作模型LoopWAM，登顶真机评测榜单RoboChallenge Table30 V2，成功率与过程分双双位列榜首，成绩超过VLA-DM0.5等同期方案。”
Exact source · revision 1jiqizhixin
Evidence dependency changes (6)
- 2026-10-30T23:04:55.343Z · evidence updated · source revision 2. Evidence extraction was updated; current source attributions are shown above.
- 2026-10-30T23:04:48.453Z · evidence updated · source revision 2. Evidence extraction was updated; current source attributions are shown above.
- 2026-10-30T22:13:13.932Z · evidence updated · source revision 1. Evidence extraction was updated; current source attributions are shown above.
- 2026-10-30T22:13:04.994Z · evidence updated · source revision 1. Evidence extraction was updated; current source attributions are shown above.
- 2026-10-16T11:34:08.253Z · evidence updated · source revision 2. Evidence extraction was updated; current source attributions are shown above.
- 2026-10-15T19:09:04.080Z · evidence updated · source revision 1. Evidence extraction was updated; current source attributions are shown above.
Report timeline
Follow attributed reports and material updates.
- Machine Heart — Robotics on WeChatSignalLoopWAM: Closed-Loop Generalist Policy Improvement from Deployment Rollouts
LoopWAM, a world action model developed by Tsinghua University, Yuanli Lingji, and Jihuan Shidai, achieves top performance on RoboChallenge Table30 V2 and other benchmarks. It leverages real-world deployment data to iteratively improve robot policies through quality-guided learning, incorporating success, suboptimal, and failure trajectories to enhance world modeling and action generation. The model uses a video-action dual-branch architecture and quality steering to guide policy improvements, achieving significant performance gains across multiple tasks.
- arXiv Robotics — research abstractsSignalLoopWAM: Closed-Loop Generalist Policy Improvement from Deployment Rollouts
This paper presents LoopWAM, a closed-loop world-action model that learns directly from full-quality deployment rollouts, including successes, suboptimal behavior, and failures. By integrating source-dependent noise sampling, quality steering, and quality-weighted supervision, LoopWAM achieves significant improvements in success rates on benchmarks like RoboChallenge Table30-V2, RoboTwin 2.0, and RoboCasa365, demonstrating effective closed-loop policy improvement under both multi-task learning and single-task post-training scenarios.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).