Home / Platforms / Bimanual arms

Platform 02

Bimanual arms

Dual-arm teleoperation rigs for bimanual manipulation, contact-rich tasks, and dexterous policies.

14
Bimanual rigs supported
30Hz
Joint logging
<8ms
Bilateral latency
BIMANUAL RIGTASKbilateral teleop14 rigs activeLEFT ARM DATARIGHT ARM DATASYNC TIMESTAMPTASK TYPES SUPPORTEDAssemblypeg-in-hole • connector matingCloth Foldinggarments • soft goodsPouring / Handoverliquids • tool transfer14bimanual rigs30Hzjoint logging<8msbilateral latency

What is bimanual arm data collection?

Bimanual platforms like ALOHA enable two-arm coordinated manipulation — the hardest class of tabletop manipulation tasks. Both arms must move in precise coordination, mirroring human two-handed dexterity.

Use cases we support:

  • Assembly — peg-in-hole, connector mating, multi-part joining
  • Cloth folding — garments, soft goods, deformable object manipulation
  • Tool handover — passing objects between hands mid-task
  • Bimanual grasping — two-arm lifts, stabilize-and-operate tasks

Why teams partner with us:

  • 14 bimanual rigs purpose-built for two-arm demos
  • 30Hz joint logging across both arms simultaneously
  • <8ms bilateral latency — left-right sync without drift
  • Operators trained on two-arm coordination, not repurposed single-arm teams

Bimanual coordination is where most tabletop policies fail.

Our rigs are purpose-built for two-arm demos — not adapted single-arm setups. That difference shows in data quality, sync accuracy, and task success rates.


14 bimanual rigs
30Hz joint logging
<8ms bilateral latency

Models we support

Aloha, dual-arm, and custom bimanual rigs

Production ALOHA variants and dual-cobot configurations across the field.

ALOHA

Stanford

ALOHA Stationary

Bench

ALOHA Unleashed

Mobile

UR3 + UR3

Dual cobot

Franka dual

Research

Custom dual

BYO config

What we capture

What we capture on a bimanual rig

The streams a bimanual VLA or contact-rich policy actually trains on.

Bimanual coordination

Synchronized left and right arm trajectories, joint and end-effector.

Tool-tip pose pairs

Per-arm TCP pose in your robot frame, calibrated per session.

Contact force profiles

Wrench data from each arm, time-aligned for two-handed contact-rich tasks.

Synchronized video

Wrist cameras on both arms plus scene cameras, frame-perfect.

How we integrate

From dual rig to production

A four-stage integration designed around bilateral teleop latency budgets.

1Step 1

Rig calibration

Both arms calibrated together. Bilateral latency measured and locked.

2Step 2

Bimanual training

Operators trained on coordination patterns specific to your task.

3Step 3

Production capture

Trajectories captured with daily QA on coordination quality.

4Step 4

Force and video sync

Force sensor and multi-camera streams aligned per session.

What our partners say
The ALOHA pipeline took us months to set up internally. Roborax had a calibrated rig and trained operators in two weeks.
Olamide Ade
Bimanual Lead, Pavlon Research

FAQ

Questions about bimanual arm data collection

Coordination. Both arms must act as a coherent system — one arm stabilises while the other manipulates, or both execute a shared force closure. This requires operators who can coordinate both controllers simultaneously and requires tighter QA on spatial consistency.
Operators train on bimanual rigs with mirrored controller setups and a split-screen view of each arm. We run coordination drills before any program data is collected.
Assembly, kitting, food preparation, folding, and tool use tasks requiring two-handed coordination. If your task requires it, we can build a custom scene with the exact objects and constraints your robot will face in deployment.

Spin up a bimanual program

Tell us the rig and the coordination pattern. Two-week ramp to first batch.