arXiv cs.AI· Octi Zhang, Iretiayo Akinola, Rosario Scalise, Mateo Guaman Castro, Patrick Yin, Tobias Widmer, Miles Macklin, Chris Amevor, Dylan Turpin, Dirk Van Gelder, Jan Carius, Ankur Handa, Gavriel State, Yunrong Guo, Yashraj Narang·· 6 天前AI 评分37
AlphaNIST：从二元任务奖励中涌现的装配技能
AlphaNIST: Emergent Assembly Skills from Binary Task Rewards
AI 导读
AlphaNIST 仅用稀疏二元任务奖励，在仿真中从零训练单一策略完成 NIST Assembly Task Board #1 全部 20 项装配任务，整体训练成功率达 98.2%，无需演示、进度塑形、脚本化原语或逐任务调参。该框架还能让策略在单回合内完成十组装配对，且仅在十组同时完成时给予奖励，策略自发涌现出为后续装配选择抓取、借助周围几何重新定向零件等灵巧且类规划的行为。
正文 · 原文
Authors:Octi Zhang, Iretiayo Akinola, Rosario Scalise, Mateo Guaman Castro, Patrick Yin, Tobias Widmer, Miles Macklin, Chris Amevor, Dylan Turpin, Dirk Van Gelder, Jan Carius, Ankur Handa, Gavriel State, Yunrong Guo, Yashraj Narang
Abstract:The NIST Assembly Task Board #1 remains a challenging benchmark for learning-based methods in robotic manipulation. Over short horizons, it demands precise, contact-rich control of small, tight-fitting parts. Over long horizons, it requires recovery and coordination across extended action sequences. Demonstrations, task decomposition, and dense progress rewards have enabled research advances by providing guidance on how to assemble. But how much of the strategy must be specified? The progression from AlphaGo to AlphaGo Zero showed that RL at scale can discover superhuman strategies without prescribing how to win. AlphaNIST asks whether a sparse binary task reward and task diversity admitted by this task-agnostic reward can similarly let a robot discover how to assemble. In simulation, we train a single policy from scratch across all 20 assembly tasks on the NIST Assembly Task Board #1, reaching 98.2% overall training success without demonstrations, progress shaping, scripted primitives, or per-task tuning. The framework also trains policies to complete ten assembly pairs in a single episode without a prescribed order, with task success rewarded only when all ten pairs are simultaneously complete. The learned policies discover dexterous and planning-like behaviors: choosing grasps for later assembly, using surrounding geometry to reorient parts, and preparing one part, assembling another, then returning to finish the first without an explicit planner. For videos, see our website: this https URL
| Comments: | 8 pages, 9 figures, ICRA |
| Subjects: | Artificial Intelligence (cs.AI); Robotics (cs.RO) |
| MSC classes: | 68T40, 68T05 |
| ACM classes: | I.2.9; I.2.6; I.2.8 |
| Cite as: | arXiv:2610.13354 [cs.AI] |
| (or arXiv:2610.13354v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.13354 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zhengyu Zhang [view email]
[v1]
Fri, 9 Oct 2026 10:29:21 UTC (26,645 KB)
来源：arXiv cs.AI · arxiv.org