Overview

ALOHA 2 is a next-generation, low-cost platform for bimanual teleoperation. It combines high-precision control with an affordable, open-source design to enable complex manipulation and data-driven research in robotics and automation.

ALOHA 2: Advancing Low-Cost Bimanual Teleoperation

ALOHA 2 is designed for high-precision two-arm teleoperation with a focus on scalability, ergonomics and flexible integration into different environments—from research labs to industrial workcells.

The system’s mechanical design, gravity compensation, cameras and grippers are optimized for:

• Precise, high-dimensional manipulation tasks using two coordinated robotic arms

• Efficient data collection for learning-based methods

• Long-duration teleoperation with reduced physical strain on the operator

Because the platform is open-source, both hardware and software can be adapted, extended and improved by the wider robotics community.

Operating Principles

The operation of ALOHA 2 combines:

• Carefully engineered hardware

• Accurate simulation models

• Software interfaces for control, teleoperation and learning

This makes it possible to:

• Adapt to a wide variety of manipulation tasks

• Collect large-scale datasets

• Train and validate policies in both the real world and simulation

Key Components

• Redesigned grippers

  • Two gripper types:

    • Leader gripper: used directly by the human operator

    • Follower grippers: mounted on the robot arms

• Emphasis on:

  • Low latency

  • Good force control

  • Ability to handle objects of different shapes, sizes and weights

• Provides intuitive operator control with precise feedback on the robot side

• Frame redesign - Aluminum extrusion structure for high stiffness and low weight - Supports complex setups, larger objects and human–robot collaboration - Easy to modify and expand to suit specific experiments

• High-resolution cameras

  • Intel RealSense D405 sensors give:

    • Dense depth information

    • Wide field of view

    • Accurate real-time 3D perception

  • Enables fine manipulation, object tracking and robust teleoperation in visually complex scenes

• MuJoCo simulation

  • A detailed MuJoCo model of the ALOHA 2 system reproduces the real robot’s kinematics and dynamics

  • Allows:

    • Policy training in simulation

    • Safety testing of algorithms

    • Rapid iteration on control and learning methods before deployment

  • The model is available in the MuJoCo ALOHA 2 repository.

Using simulation reduces wear on the physical hardware and speeds up development cycles.

ALOHA 2 targets researchers and developers who need:

• Reliable, repeatable bimanual teleoperation

• Open and modifiable hardware and software

• A platform suitable for both research and applied industrial scenarios

Advantages

• Enhanced performance - High-quality grippers with low-friction guides - Advanced gravity compensation for stable, precise motion - High-resolution depth cameras for accurate, wide-area perception

• Operator-focused design - Ergonomic handles and motion paths - Intuitive teleoperation behavior - Suitable for long data-collection sessions with reduced fatigue

• Robustness and scalability

  • Simplified but durable mechanical design

  • Low maintenance and reduced downtime

  • Modular structure:

    • Supports small lab setups

    • Scales to larger robotic workcells or testbeds

Because the system is open-source, users can adapt the architecture, share extensions and collaboratively evolve the platform over time.

Applications

ALOHA 2 is applicable in many areas where precise two-arm manipulation and teleoperation are valuable.

1. Robotic Learning

• Enables collection of large, high-quality datasets of manipulation trajectories

• Accurate grippers and reliable camera views provide rich training data for: - Object recognition - Grasping and placing - Sequential task execution and planning

2. Human–Robot Interaction (HRI)

• Ergonomic operator station and adaptable frame make it suitable for shared workspaces

• Use cases include: - Assembly assistance - Inspection tasks - Assistive and service robotics

3. Teleoperation

• Designed for responsive, high-fidelity remote control

• Potential scenarios: - Surgical or medical teleoperation - Disaster response - Handling of hazardous or sensitive materials

• Intuitive controls help operators execute complex tasks even in high-risk environments.

4. Simulation Research

• Tight integration with MuJoCo allows: - Training and testing of manipulation policies in simulation - Reproduction of complex scenarios at high fidelity

• Greatly accelerates experimentation without stressing the physical hardware.

ROS Interface

Note

ROS2 interface is available as well.

The ROS1 interface simplifies teleoperation setup by integrating:

• Dynamixel motor control

• RealSense camera streams

• Visualization tools such as RViz

Below is an overview of the typical setup.

Setup

• Install dependencies

  Run the provided script:

./dependencies.sh

• Check Dynamixel configuration

• Use Dynamixel Wizard

• Set Protocol to 2.0

• Set baud rate to 1000000 (1 Mbps)

• Limit the current of the gripper motors to avoid overload

• Configure udev rules for serial ports

• Get the serial number of each USB device:

sudo udevadm info --name=/dev/ttyUSBX --attribute-walk | grep serial

Replace /dev/ttyUSBX with the appropriate device names and then create matching udev rules so ports remain consistent.

Teleoperation

Launch the 4-arm teleoperation node:

roslaunch aloha 4arms_teleop.launch

Robot visualization

Start the RViz visualization:

roslaunch aloha_viz view_robot.launch

Depth cameras

Launch each RealSense D405 camera (sequentially):

roslaunch aloha_realsense_cameras d405_01.launch
roslaunch aloha_realsense_cameras d405_02.launch
roslaunch aloha_realsense_cameras d405_03.launch
roslaunch aloha_realsense_cameras d405_04.launch

Use Cases

The following demonstrations highlight how ALOHA 2 performs in real tasks, with synchronized camera views, RViz visualizations and real-world footage.

1. Box Unboxing A precise bimanual task where ALOHA 2 opens a box and removes its contents, demonstrating coordinated motion and contact-rich manipulation.

2. Taking Garbage with Small Brush ALOHA 2 uses a small brush to gather scattered debris, highlighting delicate surface interaction and precise tool control.

3. Removing/Cutting Wires from an Electronic Board

ALOHA 2 carries out fine manipulation on a circuit board, cutting and removing wires in confined spaces.

4. Assembling Lego-Type Organizer Boxes

ALOHA 2 assembles small containers, demonstrating accurate positioning and assembly skills in limited workspace.

5. Making Tea

ALOHA 2 prepares a cup of tea, which includes handling containers, pouring liquid and coordinating multiple steps—illustrating suitability for service robotics.

Special thanks are given to the MuJoCo ALOHA 2 project for providing key resources and contributions to this platform.