Technical features and advanced details of the SO-ARM101 robot: open-source robotics within everyone's reach

  • The SO-ARM101 is an open-source, 3D-printed robotic arm aimed at developing affordable and educational artificial intelligence.
  • It features a modular structure, a dual-arm (leader-follower), specialized motors, and high-precision sensors, ideal for simulating, training, and executing AI algorithms in physical tasks.
  • The kit is supported by an active international community, integration with the LeRobot framework, and support for advanced hardware such as NVIDIA Jetson.

so-arm101

The world of robotics is undergoing a true revolution thanks to the combination of artificial intelligence, 3D printing, and open source hardware projects. The SO-ARM101 robot is probably one of the best recent examples of how technology is enabling the creation of advanced and affordable solutions for both education and research and development. This robotic arm, developed under the umbrella of Hugging Face and its community, fuses customization capabilities with powerful AI capabilities and an open-access philosophy for all technical backgrounds.

If you're looking for a modular educational robot designed for deep learning, programming, and physical object manipulation, the SO-ARM101 may be an ideal choice. Furthermore, its extensive software and hardware ecosystem makes it one of the most flexible and educational platforms currently available for experimenting with AI and real-world robotics. Below, we provide a comprehensive analysis, integrating all the information published by leading websites and specialized stores, with technical details, kit variants, compatibility, improvements over previous generations, and available resources.

What is SO-ARM101? History and philosophy of the project

El SO-ARM101 emerges as the direct evolution of the successful SO-ARM100, a community-driven DIY robotic arm from Hugging Face and developed in collaboration with key industry players such as The Robot Studio, WowRobo, PartaBot, and Seeed Studio. The concept behind this arm is that of a Open-source 3D-printed robot designed for low-cost artificial intelligence development and training.

Thanks to 3D printing and open source design, the hardware is fully customizable and affordable, fostering access to advanced robotics even for schools, students, makers, and research environments with limited budgets. The SO-ARM101 was created with an educational and prototyping focus, allowing for both traditional learning and experimentation with cutting-edge AI algorithms.

Collaboration and community: the role of Hugging Face and its partners

so-arm101

The key to this project's success lies in the active international collaboration between companies and communities. Hugging Face is leading the development of the artificial intelligence platform and control software (LeRobot), while WowRobo and Seeed Studio are providing the physical kits and logistical support. All of these actors maintain open resources on GitHub, Discord, and forums, where any user can share progress and questions or even contribute directly to the project's development.

Main improvements compared to the SO-ARM100

  • Optimized wiring: The installation has been redesigned to avoid the disconnection problems in joint 3 that existed in the previous generation.
  • New engines: Now it incorporates improved gear ratios that do away with external gearboxes, reducing maintenance and increasing precision.
  • Real-time synchronization between the two arms: The leader-follower system allows one arm to imitate the other in real time, ideal for human reinforcement learning techniques and imitation-type training.
  • Plug-and-play support for embedded AI and edge computing, especially with NVIDIA Jetson modules and the reComputer J4021 Orin NX 16GB mini PC.

Versions, prices and composition of the SO-ARM101 kits

The SO-ARM101 is available in several versions and formats to suit different needs and budgets. There are DIY (do it yourself) kits and pre-assembled versions, as well as 'Standard' and 'Pro' variants with different engines and specifications.

  • DIY Kits: Designed for enthusiasts and makers who enjoy assembly and want to learn from scratch. It includes all the parts needed to assemble the robotic arm yourself.
  • Assembled version: Perfect for those who want to get started without complications: just unpack, plug in and start using.
  • Standard vs Pro Differences: Mainly in the type of motors (power and feed) and torque capabilities. The Pro version provides greater capacity, ideal if the kit will be used with demanding loads or applications.

The price range varies considerably depending on the distributor and the format:

  • Basic kit from approximately $100 (no 3D printed parts or assembly).
  • Complete dual kit (both arms, no printed parts): $220 (Standard)/$240 (Pro).
  • 3D printed parts: They can be purchased separately for about $35 if you don't have your own 3D printer.

It is important to note that most stores (e.g. Seeed Studio, AliExpress) 3D printed plastic parts are not included in the main kit, so the buyer must print them themselves or purchase them separately.

Dimensions, weight and shipping logistics

According to information provided by distributors, The SO-ARM101 ships in multiple packages depending on kit and accessory selection.Typical package dimensions and weights are as follows:

  • Package 1: 34cm x 29cm x 17cm; actual weight 2kg
  • Package 2: 34cm x 30cm x 25cm; actual weight 5kg
  • Package 3: 42cm x 27cm x 25cm; actual weight 5kg

International shipments are typically handled primarily through UPS, DHL, or FedEx. Some countries may have specific customs requirements, so it's a good idea to check the declared value, potential import taxes, and delivery times (in the U.S., the process is particularly expeditious and duty-free, although the time may extend to 2–4 weeks due to pre-storage in a local warehouse).

Detailed technical specifications

so-arm101 robotic arm

Getting into the technical details, The SO-ARM101 is a dual-arm robot (leader and follower), each with a 6-degree-of-freedom configuration, giving it great versatility and precision in complex movements. Let's look at its main details:

  • 6 degrees of freedom (DOF) per arm: Each arm can perform advanced movements and complex manipulations, as each joint is independently motorized.
  • Motorization:
    • Total of 12 bus-type actuators (STS3215 servos), 6 for each arm.
    • Leading arm:
      • 1 7.4V motor with 1:345 ratio for joint no. 2
      • 2 7.4V motors with 1:191 ratio for joints No. 1 and No. 3
      • 3 x 7.4V motors with a 1:145 ratio for joints #4, #5 and the gripper
    • Follower arm:
      • Standard Kit: 6 x 7.4V motors with 1:345 ratio for all joints, torque of 16.5 kg·cm at 6V.
      • Pro Kit: 6 12V motors with 1:345 ratio and 30 kg·cm torque.
  • Sensors: 12-bit magnetic encoder for each axis, providing accurate position readings.
  • interfaces: USB-C for programming and testing, communication with the host via UART.
  • Control method: Mainly through PC and the LeRobot framework, which in turn supports PyTorch-based AI models, facilitating the integration of reinforcement learning and imitation learning.
  • Feeding:
    • Standard Kit: 2 5V/4A power adapters
    • Pro Kit: 12V/1A for the follower arm and 5V/4A for the leader arm
  • Working field and temperature: It operates within a range of 0°C to 40°C.

They are also found in the official documentation and the GitHub repository Detailed assembly instructions, 3D printing files, complete component list, and calibration tutorials.

Applications, integration and available resources

The SO-ARM101 stands out for its native integration with the Hugging Face LeRobot framework, This facilitates the training and testing of AI models in tasks involving manipulation, grasping, imitation learning, and positioning of physical objects. The community has enabled various resources to get the most out of the robot:

  • Simulation and training in virtual environments before deploying models on the real robot.
  • Step-by-step tutorials for calibration and commissioning.
  • Active support is available on both GitHub and the official Discord channel, where users share progress and quickly resolve questions.
  • Tools for data collection and deployment of trained models both locally and using edge computing.
  • Support for advanced hardware solutions like the NVIDIA reComputer J4012/4021 mini PC with Jetson Orin NX 16GB, ideal for embedded AI and edge computing.
qualcomm rb3
Related article:
Qualcomm RB3 Gen 2 AI platform for robotics and IoT