Overview
DOFBOT PRO is a desktop-level 3D AI vision robotic arm designed for ROS education and development. It combines a 6-DOF motion joint structure, a 3D depth camera, and NVIDIA Jetson series control boards to simplify complex motion control through ROS, forward/inverse kinematics, and visual perception for 3D-space recognition, tracking, and grasping.
Videos
Key Features
- Jetson platform compatibility: compatible with Jetson Nano 4GB / Jetson Orin Nano SUPER / Jetson Orin NX SUPER control boards; GPU-accelerated model training and Python development are supported.
- 3D depth point cloud recognition: RGB + depth (RGB+D) fusion detection for 3D positioning, tracking, and grasping tasks.
- ROS motion planning and simulation: supports MoveIt motion planning and RViz robot simulation; supports 2D and 3D visual interaction.
- 6-DOF aluminum-alloy structure: precision-machined aluminum alloy body; high-precision servos for smooth multi-axis motion.
- Cross-platform control: supports app control (Android/iOS), wireless handle control, and PC web page control.
- Multimodal / large-model concepts (as provided): Large Language Model, Large Speech Model, Large Visual Model; includes Scalable RAG Knowledge Base and “Dual-Modal Dynamic Feedback Reasoning Architecture” descriptions.
- Algorithm frameworks listed: inverse kinematics algorithm, YOLOv11, OpenCV, MediaPipe.
For product selection and technical support, contact https://rcdrone.top/ or email support@rcdrone.top.
Specifications
DOFBOT-PRO (robotic arm system)
| Master control | Jetson Nano B01 / Jetson Orin Nano SUPER / Jetson Orin NX SUPER |
|---|---|
| Degree of freedom | 6 |
| Arm span | 350mm |
| Gripper open-close | 6cm |
| Repeatable positioning accuracy | ±0.5mm |
| Structure type | Traditional robotic arm structure |
| Camera | DABAI DCW2 Depth camera |
| Visual dimension | 3D image with depth distance information |
| Voice | AI large model voice module + speaker |
| Display | 10.1-inch display |
| Function | Interconnection control; MoveIt motion planning; RViz robot simulation; 2D visual interaction; 3D visual interaction; AI large model |
| Positioning (as described) | Embedded AI / AI large model / 3D depth visual robotic arm |
ROS Robotic Arm Configurations (as listed)
| Version | Standard Version | Ultimate Version |
|---|---|---|
| Control boards | Jetson Nano B01; Jetson Orin Nano SUPER 4GB/8GB; Jetson Orin Nano SUPER 8GB/16GB | Jetson Nano B01; Jetson Orin Nano SUPER 4GB/8GB; Jetson Orin Nano SUPER 8GB/16GB |
| Voice Module | All versions include AI large model voice module | |
| Depth Camera | DABAI DCW2 Depth Camera | |
| Display | / | HD 10.1-inch touch screen |
Controller Selection Recommendations (Jetson board specs shown)
| Item | Jetson Nano B01 4GB | Jetson Orin Nano SUPER 4GB | Jetson Orin Nano SUPER 8GB | Jetson Orin NX SUPER 8GB | Jetson Orin NX SUPER 16GB |
|---|---|---|---|---|---|
| Computing power | 0.5TFLOPS (FP16) | 34 TOPS | 67 TOPS | 117 TOPS | 157 TOPS |
| CPU | Quad-core Arm Cortex-A57 MPCore processor | 6-core Arm Cortex-A78AE v8.2 64-bit CPU; 1.5MB L2 + 4MB L3 | 6-core Arm Cortex-A78AE v8.2 64-bit CPU; 1.5MB L2 + 4MB L3 | 6-core NVIDIA Arm Cortex A78AE v8.2 64-bit CPU; 1.5MB L2 + 4MB L3 | 8-core NVIDIA Arm Cortex A78AE v8.2 64-bit CPU; 2MB L2 + 4MB L3 |
| GPU | 128-core NVIDIA Maxwell GPU | 512-core NVIDIA Ampere architecture GPU with 16 Tensor Cores | 1024-core NVIDIA Ampere architecture GPU with 32 Tensor Cores | 1024-core NVIDIA Ampere architecture GPU with 32 Tensor Cores | 1024-core NVIDIA Ampere architecture GPU with 32 Tensor Cores |
| Memory | 4GB 64-bit LPDDR4; 25.6GB/s | 4GB 64-bit LPDDR5; 51GB/s | 8GB 128-bit LPDDR5; 102GB/s | 8GB 128-bit LPDDR5; 102GB/s | 16GB 128-bit LPDDR5; 102GB/s |
| Storage | 16GB eMMC + 64GB U disk | 256GB SSD | |||
| Power | 5W - 10W | 7W , 10W , 25W | 7W , 15W , 25W | 10W , 15W , 25W , 40W | 10W , 15W , 25W , 40W |
| ROS System Version | Ubuntu18.04 + Docker + ROS2 Humble | Ubuntu22.04 LTS + ROS2 Humble | |||
Function Operation Difference (measured results shown)
| Version | Jetson Nano B01 4GB | Jetson Orin Nano SUPER 8GB | Jetson Orin NX SUPER 16GB |
|---|---|---|---|
| Robot startup (program start time) | 62s | 49s | 48s |
| 2D-face tracking (program start time / program running frame) | 4s / 10fps | 7s / 30fps | 7s / 30fps |
| 2D-gesture recognition grab blocks (program start time / program running frame) | 7s / 6fps | 6s / 30fps | 6s / 30fps |
| 2D-fingertip trajectory recognition (program start time / program running frame) | 10s / 5fps | 7s / 30fps | 6s / 30fps |
| MoveIt (program start time / program running frame) | 45s / 6fps | 43s / 30fps | 38s / 30fps |
| 3D-Yolo garbage recognition and sorting (program start time / program running frame) | 64s / 5fps | 9s / 30fps | 6s / 30fps |
| 3D-Mediapipe gesture machine code distance sorting (program start time / program running frame) | 9s / 6fps | 5s / 14fps | 3s / 15fps |
| 3D-tracking to grab color blocks (program start time / program running frame) | 8s / 10fps | 4s / 14fps | 2s / 15fps |
| AI large model for object sorting (program start time / program running frame) | 40s / 5fps | 25s / 30fps | 20s / 30fps |
Applications
- 3D vision detection and grasping; spatial perception; object tracking; 3D sorting
- Depth ranging (distance measurement), shape recognition, height measurement, volume measurement
- Depth vision positioning and tracking; 3D spatial tracking and grasping; 3D point cloud recognition
- AI-powered visual interaction: intelligent sorting and handling, color recognition, dynamic tracking, garbage sorting, tracking, grasping
- Multimodal workflows described: video analysis, long-command motion control, abnormal height sorting, intent inference (RAG knowledge base), KCF object tracking algorithm, YOLOv11-based recognition tasks
Example object dimensions shown for volumetric measurement demonstrations: 30*30*30mm Cube, 30*30*30mm Cylinder, 30*30*60mm Cylinder. Example distance overlays shown include 240.0mm and 190.0mm.
Manuals
Tutorial link: http://www.yahboom.net/study/DOFBOT-Pro
Details

Compare popular desktop robotic arm options at a glance, including degrees of freedom, reach, gripper range, and control platforms.

A quick spec snapshot helps choose the right model for ROS learning, simulation, and basic vision tasks.

DOFBOT-PRO combines a 6-DOF arm, RGB+D depth sensing, and Jetson compatibility for 3D perception and grasping development.

Alternative configuration details are provided for users who need a different arm structure and camera setup.

Built for ROS education and development, the kit pairs a compact 6-DOF arm with depth vision and an integrated desktop-style setup.

Designed for motion planning and perception workflows such as kinematics, target recognition, tracking, and grasping in 3D space.

Key modules cover depth perception, AI interaction concepts, and software frameworks used in common robotics pipelines.

Hardware and software highlights summarize what’s included for building vision + ROS demos and classroom experiments.

Multiple Jetson board options help scale from entry-level prototyping to higher-performance AI workloads.

Use the configuration matrix to match the controller board and feature set to your ROS project requirements.

Depth vision adds distance-aware understanding for more reliable positioning, recognition, and grasp planning than 2D alone.

Arm-camera calibration supports tasks like point-cloud recognition and depth-based measurement for 3D-space interaction.

Multimodal interaction concepts include text, voice, and vision capabilities for building richer human–robot workflows.

Application examples focus on sorting and handling behaviors that combine perception with command-driven control.

Practical demos showcase tracking, sorting, and action selection tasks built around vision and interaction logic.

Interactive challenge-style activities provide approachable scenarios for testing perception, reasoning, and control loops.

Vision recognition examples cover color-based tracking, block sorting, interactive games, and label-based stacking.

Training notes and performance curves outline the included deep-learning workflow direction for object detection tasks.

DOFBOT Pro supports MediaPipe-based gesture interaction, forward/inverse kinematics, and MoveIt simulation control for setup and development workflows.

DOFBOT Pro supports MoveIt kinematics simulation with trajectory planning, collision detection, and ROS/ROS2 (Humble) workflows for motion control.

DOFBOT Pro supports app control, web control, and a USB wireless remote, with a 6-DOF joint layout labeled J1–J6 for precise setup and movement planning.

The DOFBOT Pro 6-DOF robotic arm pairs a Jetson-based control board with a DaBai DCW2 depth camera and intelligent serial bus servos for vision-guided motion projects.

The DOFBOT Pro setup includes a robotic arm expansion board layout and supports add-ons like a voice module and a 10.1-inch touch screen for control.

The DOFBOT-PRO course outline breaks down the training modules and learning goals to help plan setup and development steps.

DOFBOT Pro includes organized open-source code and step-by-step tutorial folders covering 2D/3D visual tracking, sorting and gripping, and depth camera workflows.

DOFBOT Pro includes downloadable video tutorials, ROS2 learning materials, a 3D model file, and open-source Python code for development on Jetson boards.

Dimension drawings and a spec overview help you plan mounting space and system integration for the DOFBOT Pro 6-DOF robotic arm.

The DOFBOT Pro kit includes the robotic arm with a set of standard accessories such as controller hardware, power and data cables, and basic tools for assembly and setup.
Related Collections
