ALOHA

Stanford's open-source, low-cost bimanual teleoperation platform, priced at around $20,000.

ALOHA is an open-source bimanual (two-armed) robot hardware platform released in 2023 by Tony Zhao, Chelsea Finn, and colleagues at Stanford, with collaborators from Berkeley and Meta, alongside the paper “Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware.” A full setup costs about $20,000. It uses two small leader arms, moved by hand, to drive two matching follower arms for teleoperation data collection, along with four cameras (two fixed, two wrist-mounted). The same paper introduced the ACT algorithm, which could learn precise bimanual tasks such as threading a zip tie or inserting a battery from only about 10 minutes of demonstrations. ALOHA lowered the barrier to collecting bimanual manipulation data, and later spawned variants such as Mobile ALOHA and ALOHA 2; it has also become a common real-robot platform for many VLA papers.

ExampleBoth π0 and ACT have been demonstrated on ALOHA-style bimanual platforms performing tasks such as folding laundry and inserting batteries.

Also called

A Low-cost Open-source Hardware System for Bimanual Teleoperation

Related

Action Chunking with Transformers · Mobile ALOHA · ALOHA 2 · Leader-Follower Teleoperation · Bimanual Manipulation · Trossen Robotics ViperX 300

Sources

ALOHA project page (Tony Zhao et al.)

As of

2023-04