Developed by Stanford University, it is specifically designed to perform complex movement tasks that require coordination of both hands and the whole body.
It can autonomously complete various tasks in daily life through imitation learning (i.e., observing human operations and then imitating these actions) through joint training through only 50 task demonstrations.
Such as cooking, opening cabinets and putting things, and taking the elevator by yourself.
And most importantly: the robot’s hardware and software are all open source.
Key features:
1. Low-cost full-body remote operating system: Mobile ALOHA is an affordable system that allows users to collect data through full-body remote operation. This design makes the system more accessible and accessible.
2. Two-Hand Mobile Operation: The system focuses on mimicking mobile operation tasks that require both hands and full-body control, an ability that is often missing in traditional desktop-operated robots. Such as cooking, cleaning, or other activities that require both hands to work together.
3. Autonomous imitation learning: Mobile ALOHA utilizes imitation learning technology to train robots to perform complex tasks by observing human demonstrations. With 50 demonstrations per task, co-training can increase the success rate to 90%, enabling Mobile ALOHA to autonomously complete complex mobile operation tasks.
4. Dataset co-training: Using the data collected by Mobile ALOHA, the research team conducted supervised behavioral cloning and found that co-training with existing static ALOHA datasets can improve the performance of mobile operation tasks.
5. High success rate task execution: Through joint training and imitation learning, Mobile ALOHA can autonomously complete complex mobile operation tasks with a success rate of up to 90%.
6. Diverse Application Scenarios: Mobile ALOHA is capable of performing a variety of complex tasks, such as cooking, opening cabinet doors, operating elevators, and cleaning work, showcasing its wide application potential.
7. Compatibility with IoT devices: In addition to performing physical tasks, Mobile ALOHA plans to connect with Internet of Things (IoT) devices in 2025, further expanding its application scope.
Mobile ALOHA’s hardware composition:
The Mobile ALOHA is mounted on a mobile pedestal designed for warehouses: the Tracer AGV.
It can carry 100kg and move at speeds of up to 1.6m/s at a cost of just $7k, while reducing the robot’s footprint by 45% and reducing its weight by 15kg. The robot can reach a vertical height of 65 cm to 200 cm and a distance of 100 cm from the base.
- Robotic Arms: The Mobile ALOHA is equipped with two robotic arms, each with multiple degrees of freedom, allowing it to perform complex two-handed operation tasks.
- Mobile Base: The system includes a mobile base that enables the robot to move and position itself in different environments.
- Cameras: Mobile ALOHA is equipped with two wrist cameras and a top camera to capture visual information about the environment and operational tasks.
- Built-in power and computing power: The system has its own power supply and computing power, allowing it to complete tasks independently without relying on external power supplies or computing devices.
- Technical Specifications:
- Weight: 75 kg.
- Dimensions: 80W x 84L x 140H cm (without joystick); 90 W x 135 L x 140 H cm.
- Load capacity: 750 g per arm, 55 kg base
- Arm repeatability accuracy: 1 mm.
- Arm positioning accuracy: 5-8 mm.
- Battery life: 12 hours (1620Wh).
- Maximum pulling force: 100 Newtons, vertical distance 100 cm.
- Rolling resistance: 13 Newtons (vinyl flooring).
- Movement speed: can reach 1.42 m/s of normal human walking
Projects and Presentations: https://mobile-aloha.github.io
Learn Code: https://github.com/MarkFzp/act-plus-plus
Hardware code: https://github.com/MarkFzp/mobile-aloha
Thesis: https://mobile-aloha.github.io
Tutorial: https://docs.google.com/document/d/1_3yhWjodSNNYlpxkRCPIlvIAaQ76Nqk2wsqhnEVM6Dc