Driving · 2026-10-22
Robot Generalization Beyond Volume: The Diversity Path via Zero-Shot Capabilities
Rafael Felix
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
The authors formulate zero-shot generalization in physical AI as coverage of the deployment distribution, arguing that increasing samples from the same collectors shrinks the unseen region only slowly, while each new contributor can extend it. They propose a shared latent space for robot actions combining multi-modal alignment, generative augmentation, and noisy-label learning, plus safety signals via seen/unseen detection and out-of-distribution noise estimation. They also propose a developer community building on Origami.
Why it matters
The paper targets a core problem for general-purpose robots: succeeding on situations not covered by training. By tying generalization to diversity/coverage (not just data volume) and proposing a shared latent space with uncertainty-aware safety, it offers a concrete direction for scaling physical zero-shot capability using diverse community contributions on a standard robot platform.
Method
- Formulate zero-shot generalization in physical AI as success on unseen situations, governed by coverage of the deployment distribution and by the diversity of contributors (coverage can grow as the union of contributors’ support).
- Build a shared latent space for robot actions using multi-modal alignment (vision-language-state/action), generative augmentation to fill gaps, and noisy-label learning with uncertainty learning to absorb imperfect demonstrations.
- Close the sim-to-real gap with a Real2Sim2Real loop where each layer is developed first in simulation, and propose safety from pre-training onward using seen/unseen detection and out-of-distribution noise estimation; hand control to a human via teleoperation when in unseen situations.
Abstract (from arXiv)
General-purpose robots must handle tasks, objects and environments they were never trained on, yet zero-shot generalization remains one of the least solved problems in physical AI. In digital AI, zero-shot capability grew with the volume and, above all, the diversity of training data, much of it sourced from the crowd. Robotics instead relies on paid collection by a few well-funded teams, which yields volume without diversity. We formulate zero-shot generalization in physical AI as coverage of the deployment distribution, and show that more samples from the same collectors shrink the unseen region only slowly, while each new contributor can extend it. Building on this and on zero-shot learning, we propose a shared latent space for robot actions that combines multi-modal alignment, generative augmentation, and noisy-label learning. As a scalable source of diverse data, we propose a community of developers building applications on one standard, approachable robot, Origami. We close with a testable hypothesis and an evaluation protocol relating real-robot zero-shot success to the diversity of contributed data.
Related papers
- Imagining Goals with Geometric Horizon Models for Policy Generalization
- RoboJEPA: Scaling Robotic Latent World Models
- CASTER: Formalizing Zero-Shot Manipulation from Synthetic Videos as Constrained Optimization
- VIGOR: Zero-Shot Visual Generalization via Latent-Space Consistency in Model-Based Reinforcement Learning
- GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-21 · Indexed by Pub-AI: 2026-10-22 UTC
Robot Generalization Beyond Volume: The Diversity Path via Zero-Shot Capabilities
Preprint posted: 2026-10-21 · Original record ↗
- arxiv
- 2610.29829
Explicit source versions (1)
- v2 · 2026-10-22
- Preprint available
Source date: 2026-10-21 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
- Explicit source version detected · v2
Source date: 2026-10-22 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:id explicit version (official metadata)
Abstract · arxiv
General-purpose robots must handle tasks, objects and environments they were never trained on, yet zero-shot generalization remains one of the least solved problems in physical AI. In digital AI, zero-shot capability grew with the volume and, above all, the diversity of training data, much of it sourced from the crowd. Robotics instead relies on paid collection by a few well-funded teams, which yields volume without diversity. We formulate zero-shot generalization in physical AI as coverage of the deployment distribution, and show that more samples from the same collectors shrink the unseen region only slowly, while each new contributor can extend it. Building on this and on zero-shot learning, we propose a shared latent space for robot actions that combines multi-modal alignment, generative augmentation, and noisy-label learning. As a scalable source of diverse data, we propose a community of developers building applications on one standard, approachable robot, Origami. We close with a testable hypothesis and an evaluation protocol relating real-robot zero-shot success to the diversity of contributed data.