**ALOHA 2: An Enhanced Low-Cost Hardware for Bimanual Teleoperation** The ALOHA 2 team, comprising researchers from Google DeepMind, Stanford University, and Hoku Labs, introduces ALOHA 2, an enhanced version of the original ALOHA platform designed for bimanual teleoperation. ALOHA 2 aims to improve performance, ergonomics, and robustness while reducing hardware costs and complexity. Key improvements include: 1. **Grippers**: New low-friction rail designs for both leader and follower grippers, enhancing teleoperation ergonomics and responsiveness. 2. **Gravity Compensation**: A passive gravity compensation mechanism using off-the-shelf components, improving durability and reducing operator strain. 3. **Frame**: Simplified frame design with additional space for human-robot interaction and props, maintaining rigidity for camera mounting. 4. **Cameras**: Upgraded to smaller Intel RealSense D405 cameras with larger fields of view, depth capabilities, and global shutter, reducing the footprint of the follower arms. 5. **Simulation**: A detailed MuJoCo model of ALOHA 2 with system identification, enabling high-quality data collection and policy learning in simulation. These enhancements make it easier to collect large datasets for complex manipulation tasks, such as folding a T-shirt, tying a knot, or throwing objects. The team also provides a detailed tutorial and open-source hardware designs to facilitate research and collaboration. The project website is available at [alooha-2.github.io](https://github.com/alooha-2/alooha-2). **Conclusion** ALOHA 2 is a low-cost, high-performance system designed for bimanual teleoperation, aiming to advance research in robot learning through large-scale data collection. The team hopes that the open-source nature of ALOHA 2 will accelerate progress in fine-grained bimanual manipulation tasks.**ALOHA 2: An Enhanced Low-Cost Hardware for Bimanual Teleoperation** The ALOHA 2 team, comprising researchers from Google DeepMind, Stanford University, and Hoku Labs, introduces ALOHA 2, an enhanced version of the original ALOHA platform designed for bimanual teleoperation. ALOHA 2 aims to improve performance, ergonomics, and robustness while reducing hardware costs and complexity. Key improvements include: 1. **Grippers**: New low-friction rail designs for both leader and follower grippers, enhancing teleoperation ergonomics and responsiveness. 2. **Gravity Compensation**: A passive gravity compensation mechanism using off-the-shelf components, improving durability and reducing operator strain. 3. **Frame**: Simplified frame design with additional space for human-robot interaction and props, maintaining rigidity for camera mounting. 4. **Cameras**: Upgraded to smaller Intel RealSense D405 cameras with larger fields of view, depth capabilities, and global shutter, reducing the footprint of the follower arms. 5. **Simulation**: A detailed MuJoCo model of ALOHA 2 with system identification, enabling high-quality data collection and policy learning in simulation. These enhancements make it easier to collect large datasets for complex manipulation tasks, such as folding a T-shirt, tying a knot, or throwing objects. The team also provides a detailed tutorial and open-source hardware designs to facilitate research and collaboration. The project website is available at [alooha-2.github.io](https://github.com/alooha-2/alooha-2). **Conclusion** ALOHA 2 is a low-cost, high-performance system designed for bimanual teleoperation, aiming to advance research in robot learning through large-scale data collection. The team hopes that the open-source nature of ALOHA 2 will accelerate progress in fine-grained bimanual manipulation tasks.