arXiv cs.AI· Xin Wang, Wenhao Wu, Menghao Zhang, Zhi Wang, Kun Shao, Jian Luan, Yang Li, Qing Li, Shangding Gu, Huichi Zhou, Shuqing Shi, Fei Ni, Shuo Lu, Weicheng Meng, Kang Li, Jin Wu, Kang Zhao, Shangmin Guo, Gen Li, Yongqiang Tang, Zhizhong Zhang, Yuan Xie, Heng Qu·· 13 小时前AI 评分50
HarnessPAI：面向物理AI的可演化Harness框架
HarnessPAI: An Evolving Harness for Physical AI
AI 导读
研究团队提出HarnessPAI，一个与模型和本体无关的物理AI Harness框架，把代码作为组织底层动作原语的可执行、可演化接口。该框架在单次rollout内以程序级开环执行，跨rollout则用执行反馈修订程序并把失败蒸馏为可复用技能。在桌面机械臂、家用机器人、扫地机器人和四足行走智能体上，HarnessPAI相对基线均取得一致提升，收敛后的程序还可为VLA和WAM模型生成微调训练数据。
正文 · 原文
Authors:Xin Wang, Wenhao Wu, Menghao Zhang, Zhi Wang, Kun Shao, Jian Luan, Yang Li, Qing Li, Shangding Gu, Huichi Zhou, Shuqing Shi, Fei Ni, Shuo Lu, Weicheng Meng, Kang Li, Jin Wu, Kang Zhao, Shangmin Guo, Gen Li, Yongqiang Tang, Zhizhong Zhang, Yuan Xie, Heng Qu
Abstract:Physical AI aims to build embodied agents that perceive the world, understand and reason about it, and decide how to act. Yet the field has focused primarily on the last component: the action model that maps observations to low-level controls. The prevailing training recipe can erode the perceptual and reasoning capabilities needed for robust behavior, leaving even strong action models vulnerable to scene perturbations and long-horizon tasks. We introduce HarnessPAI, a model- and embodiment-agnostic Harness framework for Physical AI that treats code as the executable and evolvable interface that organizes the underlying action primitive. The framework separates two timescales: within a rollout, it executes open-loop at the program level, with a fixed program guiding and checking execution; across rollouts, it evolves closed-loop, using execution feedback to revise the program and distill failures into reusable skills. Across desktop robot arms, household robots, a robot vacuum, and a legged walking agent, HarnessPAI demonstrates consistent improvements against baselines. Beyond the harness, the converged program also acts as a low-cost, reliable source of expert data: it generates training data for fine-tuning VLA and WAM models on tasks where it is hard for these models to obtain successful trajectories on their own. Our results suggest that the frontier of Physical AI depends not only on stronger action models, but also on executable harnesses that integrate perception, task understanding and reasoning, and action execution into a unified, verifiable, and feedback-driven system. Website: this https URL
| Comments: | 45 pages, 24 figures, 15 tables |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2609.29166 [cs.RO] |
| (or arXiv:2609.29166v2 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2609.29166 arXiv-issued DOI via DataCite |
Submission history
From: Wenhao Wu [view email]
[v1]
Thu, 24 Sep 2026 07:40:00 UTC (11,779 KB)
[v2]
Wed, 21 Oct 2026 03:58:49 UTC (8,949 KB)
来源：arXiv cs.AI · arxiv.org