Markdown edition

# IORA uses tabletop support for coupled robot hands

> IORA plans supported grasps; physical trials remain limited.

By BIGCHANGE.AI Editorial

Published: 2026-10-25T07:58:50.634Z
Updated: 2026-10-25T07:58:50.634Z
Canonical: https://bigchange.ai/blog/iora-tabletop-contact-coupled-robot-hand-grasps

![Mounted robot hand touching an orange object on a table.](https://bigchange.ai/api/media/file/iora-supported-contact-hero-v1.png)
AI-generated conceptual illustration by BIG CHANGE; not a depiction of the authors' hardware or experiment.

The [October 22 IORA preprint](https://arxiv.org/html/2610.30969v1) describes planning grasps for coupled hands.

## The big change

- **What changed:** IORA plans a stationary hold with load-bearing table contact.
- **Why it matters:** Engineers gain a contact source when planning for coupled joints.
- **What to watch:** Planning requires known object geometry and a static support surface.

## How IORA selects a hold

IORA matches hand and object regions under coupled motor constraints, refines the grasp, then filters candidates in force simulations. Table friction contributes.

## What the evidence shows

In the authors' 200-object MuJoCo test, IORA succeeded on 88.5% of objects, Dexonomy on 42.5%. Per-grasp ranking reverses because failed generation can be excluded.

Physical tests used 24 replicas. A pulley trial executed 17 of 18 grasps. With six real vegetables, the hand held against a table during **teleoperated** knife or slicer use. Autonomous cooking was untested.

## Sources

- [IORA paper, version 1](https://arxiv.org/html/2610.30969v1) — Author-reported methods and experiments; CC BY 4.0.