# AlphaNIST: Emergent Assembly Skills from Binary Task Rewards

arXiv: https://arxiv.org/abs/2610.13354 · HTML: https://mlsanity.com/paper/2610.13354/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 21 of its 28 references that are in the map.
- First posted on arXiv 9 Oct 2026.

## Builds on (most cited first, 21 of 21)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Gradient Surgery for Multi-Task Learning](https://mlsanity.com/paper/2001.06782/) (2020)
- [Learning Dexterous In-Hand Manipulation](https://mlsanity.com/paper/1808.00177/) (2018)
- [Reverse Curriculum Generation for Reinforcement Learning](https://mlsanity.com/paper/1707.05300/) (2017)
- [DeXtreme: Transfer of Agile In-hand Manipulation from Simulation to Reality](https://mlsanity.com/paper/2210.13702/) (2022)
- [Multi-task Deep Reinforcement Learning with PopArt](https://mlsanity.com/paper/1809.04474/) (2018)
- [General In-Hand Object Rotation with Vision and Touch](https://mlsanity.com/paper/2309.09979/) (2023)
- [SimBa: Simplicity Bias for Scaling Up Parameters in Deep Reinforcement Learning](https://mlsanity.com/paper/2410.09754/) (2024)
- [In Defense of the Unitary Scalarization for Deep Multi-Task Learning](https://mlsanity.com/paper/2201.04122/) (2022)
- [Deep Reinforcement Learning for Industrial Insertion Tasks with Visual Inputs and Natural Rewards](https://mlsanity.com/paper/1906.05841/) (2019)
- [Factory: Fast Contact for Robotic Assembly](https://mlsanity.com/paper/2205.03532/) (2022)
- [IndustReal: Transferring Contact-Rich Assembly Tasks from Simulation to Reality](https://mlsanity.com/paper/2305.17110/) (2023)
- [DextrAH-RGB: Visuomotor Policies to Grasp Anything with Dexterous Hands](https://mlsanity.com/paper/2412.01791/) (2024)
- [Robust Multi-Modal Policies for Industrial Assembly via Reinforcement Learning and Demonstrations: A Large-Scale Study](https://mlsanity.com/paper/2103.11512/) (2021)
- [AutoMate: Specialist and Generalist Assembly Policies over Diverse Geometries](https://mlsanity.com/paper/2407.08028/) (2024)
- [mjlab: A Lightweight Framework for GPU-Accelerated Robot Learning](https://mlsanity.com/paper/2601.22074/) (2026)
- [ARCH: Hierarchical Hybrid Learning for Long-Horizon Contact-Rich Robotic Assembly](https://mlsanity.com/paper/2409.16451/) (2024)
- [Emergent Dexterity via Diverse Resets and Large-Scale Reinforcement Learning](https://mlsanity.com/paper/2603.15789/) (2026)
- [Benchmarking Off-The-Shelf Solutions to Robotic Assembly Tasks](https://mlsanity.com/paper/2103.05140/) (2021)
- [Towards Generalized Robot Assembly through Compliance-Enabled Contact Formations](https://mlsanity.com/paper/2303.05565/) (2023)
- [Refinery: Active Fine-tuning and Deployment-time Optimization for Contact-Rich Policies](https://mlsanity.com/paper/2510.11019/) (2025)

## Abstract (arXiv)

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: https://alpha-nist.github.io/