Waveshare 5 + 1 DOF High-Torque Serial Bus Servo Dual Robotic Arm AI Kit

The RoArm-M3 is a 5 + 1 DOF high-torque serial bus servo dual robotic arm AI kit, supporting the Hugging Face LeRobot AI Robotic Project. Based on ESP32, it features a 2-DOF wrist joint, flexible expansion capabilities, secondary development support, and wireless control.

Kit Selection Guide

Model Selection Guide

Need stronger AI inference performance? Choose the ORIN-AI-Kit Series:

• Options for Jetson Orin Nano Super or Jetson Orin NX Super (Waveshare 8GB kit) to achieve low-latency, high-throughput AI computing.
• The Orin platform offers rich interfaces and powerful computing capabilities, enabling robotic arms to perform better in visual recognition, path planning, and real-time feedback, significantly enhancing the overall scalability and intelligence of the system.

Or choose the RoArm-M3-S/Pro-AI-Kit Version if you have a Jetson Orin development board or PC:

• More flexible to use with your own controller: The kit includes all necessary accessories except the main control board, allowing direct use with your PC or existing Jetson Orin development board.
• More cost-effective: Retains full functionality while reducing entry costs for robotic arm applications, making development easier and more flexible.

RoArm-M3-S / RoArm-M3-Pro robotic arm options. Choose Pro version for higher stability and precision:

• The RoArm-M3-Pro features a more robust mechanical structure and higher joint precision compared to the RoArm-M3-S version, making it suitable for industrial or research scenarios with repeatability and strict requirements of anti-interference capabilities.

Parameter Comparison

Not including the weight of the table edge fixing clamp (290±10g).

RoArm-M3-AI-Kit: 5 + 1 DOF Robotic Arm AI Kit

Supports the popular Hugging Face LeRobot AI robotics project, comes with stable and regularly updated demos, suitable for use with Jetson Orin Nano Super / Jetson Orin NX Super or a PC.

Product Introduction

The RoArm-M3-AI-Kit series supports the popular AI robotic arm project LeRobot, integrating its pretrained models and datasets with demonstrations, providing strong support for research directions such as Imitation Learning and Reinforcement Learning. By utilizing the demos and algorithm libraries of LeRobot, developers can quickly deploy intelligent algorithms onto the RoArm-M3, further expanding the possibilities for innovative applications.

RoArm-M3-S/Pro in the kit is a 5 + 1 DOF smart robotic arm designed for innovative applications. It adopts a lightweight structure design with an effective payload of 0.2kg@0.5m. It can be flexibly mounted on various mobile platforms. It adopts a 360° omnidirectional base combined with five flexible joints to create a workspace with 1-meter diameter.

The joint direct-drive design enhances repositioning precision and also improves structural reliability, with innovative dual-drive technology doubling the shoulder joint torque. The onboard ESP32 MCU main control module supports multiple wireless control modes and provides control interfaces and rich communication protocols for easily connecting to various devices.

It provides a user-friendly and cross-platform WEB application that integrates a simple and visualized coordinate control mode, making it easier to get started. It comes with rich graphic and video tutorials to help you learn and use it quickly. Compatible with ROS2 and various host computers, it supports various wireless and wired communication modes. It comes with expansion plates, supporting customizing the EoAT (End of Arm Tooling) to meet innovative application requirements.

The RoArm-M3-AI-Kit series achieves an excellent balance between lightweight, user-friendliness, expandability, and open innovation. It is a multi-functional robotic arm that integrates intelligent control, human-machine interaction, and customizable development. Ideal for applications that require a combination of flexibility, expandability, and user-friendliness.

Compatible with the popular AI robotic arm project - LeRobot

The RoArm-M3 robotic arm series supports the open-source LeRobot framework on Hugging Face, providing core resources such as pretrained models, datasets with human collected demonstrations, and simulation environments. When combined with the N-VIDIA Jetson Orin Nano Super / Jetson Orin NX Super motherboard or a PC with an N-VIDIA discrete GPU, users can experience the complete process of Imitation Learning and Reinforcement Learning projects.

LeRobot RoArm-M3 Robotic Arm Kit

Adopts a Leader-Follower robotic arm solution to collect data and build datasets for subsequent model training in imitation learning and reinforcement learning.

Flexible Joints and Omnidirectional Space

Adopts a 360° rotation base and flexible joints to create an omnidirectional operating space with a 1-meter diameter, allowing the Robotic Arm to move freely in all directions.

2 DOF Wrist Joint + EoAT for More Flexible Clamping

Equipped with a 2 DOF wrist joint with pitch + horizontal rotation support, enabling seamless coordination with the EoAT for multi-dimensional clamping and operation.

Easy to Control via Cross-platform Web Application

No App installation required; access after entering the address. Allows users to connect and control RoArm-M3 via mobile phones, tablets, and computers by clicking the buttons on the Web App. Supports secondary development of this open-source Web App to customize the user interface and add new functions.

Completely Open Source for Secondary Development

Completely open source for the control codes and communication interface documents of the RoArm-M3 series, supporting multiple languages and devices for secondary development. Provides modularized demos and tutorials for zero-based users, making it easy to get started.

Joint Direct-Drive Design: Feedback Precision Up to 0.088°

The joint angle feedback can be obtained directly via the 12-bit high-precision magnetic encoder without any reduction groups, which is more accurate. The actual position of the current target point can be calculated based on the joint angle feedback.

Innovative Dual-Drive Technology Doubles Shoulder Joint Torque

We have developed a dual-drive control algorithm that allows the two servos at the shoulder joint to coordinate their output torque effectively, significantly enhancing the power and the overall load capacity of the RoArm-M3 Series.

Direct-Drive Clamp Enables Precise Force Control

The direct-drive clamp design allows precise control of the clamping force in the program, suitable for clamping delicate objects without applying excessive force.

Dynamic External Force Adaptive Control

After enabling this function, you can limit the maximum torque for each joint. When the external force applied to the joint exceeds the torque threshold (configurable), the robotic arm will rotate in response to the external force and will return to the specified position when the external force decreases below the torque threshold.

One-Click Automatic Recording: Easy to Create Actions

RoArm-M3 can save JSON instructions for robot control as task files. These task files can be saved in the Flash of ESP32, which will not be lost in case of power loss. The robotic arm can perform complex and repetitive operations by calling these task files.

Accurately Reach the Target by Entering the Coordinates

Open source for the inverse kinematics control algorithm in the three-dimensional Cartesian coordinate system. After entering the target position, the robot arm can accurately reach the target point by using the inverse kinematics function to calculate the rotation angle of each joint.

Curve Velocity Control for Smoother Movement

We use the curve velocity control algorithm to make the robotic arm move smoother and more naturally, without oscillations during start and stop.

ESP-NOW Wireless Control: Low-Delay Remote Collaboration

ESP-NOW is a low-delay ad hoc wireless communication protocol that doesn't require network infrastructures. You can use one robotic arm to wirelessly control other robotic arms, and the control methods can be set as broadcast control, group control, and one-to-one control.

Note: The broadcast control has no limitation for the number of controlled devices, while the group control allows up to 20 devices. Please refer to the WIKI for more details.

Supports Leading-Following Control Mode Based on ESP-NOW

In the Leading-following control mode, the Leader robotic arm will send its joint angle information to other robotic arms via ESP-NOW communication. The other robotic arms which are in ESP-NOW Follower mode will imitate the same actions as the Leader in real time.

Multiple Control Interfaces, Devices, and Languages Support

The RoArm-M3 series can be controlled not only by the WEB application but also supports using other controllers to send JSON commands via HTTP protocol network requests, serial communication, or USB communication for controlling and obtaining feedback.

Multiple Installation Methods for Integrating into Your Application

Provides multiple installation schemes and related secondary development resources, suitable for different usage scenarios and can be flexibly integrated into your projects and applications.

Easy to Install Different Peripherals

The main structure of the upper arm adopts two 1020 European standard aluminum rails, supporting the installation of additional peripherals via boat nuts to meet different needs. It comes with boat nuts and corresponding M4 screws for easy installation.

Supports Installing a Camera to Extend Visual Applications

Comes with a camera holder and open source for the related drawings. You can directly install a camera with the corresponding size or other cameras with an adapter.

For reference only; only the LED light is included.

Onboard ESP32 Main Control Board

Adopts the powerful ESP32 main control MCU, supporting multiple wireless communication protocols, with a large amount of open source resources to help secondary development of innovative applications.

Compatible with ROS2, Provides Model Establishment File and Supports Multi-Device Cooperation

Provides communication nodes of ROS2, URDF model description file, and a zero-based tutorial for ROS2.

Rich Tutorial Resources for Secondary Development

We provide complete unit tutorials and cases, including various functions and common application scenarios, to help users get started quickly for secondary development.

General Driver Board for Robots

Onboard rich interfaces and resources for innovative development and functional expansion.

Diversified Onboard Resources Supports Function Expansion

Integrated with INA219 battery voltage monitoring and 9-axis IMU sensor, supporting function expansion and innovation.

Mouse Drag-and-Drop Interaction

Provides Web applications and desktop software (open-source programs developed by Python), supporting setting action instructions when the mouse is pressed and released to control the robotic arm flexibly.

Specifications

Product & Packaging Show

Dimensions