No-Code Desktop Cleaning Robotic Arm - SO-ARM101 Low-Cost AI Arm Servo Motor Kit - dumpling—li - CodeCraft Hub

Deploy steps

1. Hardware Assembly and Environment Setup

   Before starting training, please complete the robotic arm kit preparation, hardware assembly, and software installation to ensure that device connection, data collection, and cloud training can be completed properly. This tutorial uses a Windows computer as an example. macOS computers can also be configured using a similar workflow. The actual serial port names and system interface may vary depending on your device.

   • Purchase the robotic arm kit: This project uses the SO-ARM101 Arm Kit as an example. You can purchase it from the Seeed Studio official store: https://www.seeedstudio.com/SO-ARM101-Low-Cost-AI-Arm-Kit-Pro-p-6427.html . Please confirm the kit version and included accessories before purchasing.

   • Assemble the robotic arm: Follow the official assembly video to install the Leader arm and Follower arm: https://www.youtube.com/watch?v=M_ue0wcX3Qs . After assembly, check that the servos, cables, power connections, and mechanical structure are firmly connected.

   • Install the training software: This tutorial uses the SenseCraft Robotics graphical software for no-code training（ https://sensecraft.seeed.cc/robotics ）, so there is no need to manually configure Python or the LeRobot…

2. Software Setup Steps

   Home Page Introduction

   After entering the SenseCraft Robotics home page, you can see the complete robotic arm training workflow provided by the platform, including Device setup, Data collection, Dataset, Cloud training, and Model run. Click Get started to start configuring the device and creating a training task step by step.

   1.Settings

   1.1 Device Pairing

   After entering the Settings page, the platform guides users through the device configuration required before robotic arm training. This page includes 5 steps: Device pairing, Bind serial, Bind camera, Calibration check, and Teleop check. These steps are used to confirm the robotic arm model, bind device connections, configure cameras, and check whether the robotic arm can be calibrated and teleoperated normally.

   First, enter Device pairing. In the Select device model section, select the robotic arm model used in this project: SO100 / SO101. This model corresponds to the SO100 / SO101 Leader + SO100 / SO101 Follower combination, which is suitable for a training workflow where one Leader arm is used for demonstration and one Follower arm executes the motion. After confirming the selection, click Next to enter Bind serial.

   1.2 Bind Serial

   …

3. Verify the Trained Model

   After completing Run, observe whether the robotic arm can automatically recognize the paper trash on the desktop and place it into the specified shallow box. If the robotic arm can complete the action stably, the training workflow is complete. If grasping fails, the motion is offset, or the object cannot be accurately placed into the box, you can return to the Collection stage to add more high-quality demonstration data, delete incorrect data, and retrain the model.

   Through this tutorial, you have completed the full robotic arm training workflow from Settings, Collection, Dataset, Training, to Run.

   SenseCraft Robotics supports no-code training, making it suitable for quickly validating robotic arm automation tasks and expanding to more applications based on different objects, positions, and operation workflows.