Introduction
DOFBOT PRO is a desktop-level 3D AI vision robotic arm. Equipped with a series of high-performance hardware such as 3D depth camera, 6-DOF joint, NVIDIA Jetson series board, 10.1-inch touch screen, etc. By integrating ROS robot operating system and forward/inverse kinematics algorithms, the complex motion control of the 6-DOF robotic arm is simplified. We integrate 3D vision technology into robotic arm control to realize depth ranging, shape recognition, height measurement, volume calculation and other functions. Based on these data,DOFBOT PRO can accurately identify, track and grab objects in 3D space. Yahboom provides professional ROS course documents and functional source codes. Whether it is ROS education scenario or technology research and development, this robot can provide higher programmability and computing power, which is an ideal choice.
Features
- Based on NVIDIA Jetson series boards, Higher computing power, smoother operation
DOFBOT PRO compatible with Jetson NANO 4GB/Jetson Orin NANO/Jetson Orin NX board. The machine learning model is trained through the GPU of Jetson board. Through python programming, adopts deep learning framework, and is jointly developed with MediaPipe to realize AI creative projects such as object image recognition, gesture control, and human posture recognition.
- Depth point cloud recognition, gripping in 3D space
Combined with a high-performance 3D depth camera, DOFBOT-PRO can not only realize environmental perception, object recognition and other functions. It can also perform precise 3D space positioning based on the target RGB data, position coordinates and depth information, combined with RGB+D fusion detection, to complete complex visual recognition and grasping tasks.
- 6 DOF structural design design, aluminum alloy material
The entire robotic arm body is made of high-quality aluminum alloy and has six flexible degrees of freedom. Each joint uses a high-precision servo, with faster response speed, can perform smooth multi-axis movements.
- Multiple cross-platform remote control methods
DOFBOT PRO supports APP (Android/iOS), wireless handle, and PC web page control, allowing you to control the robot as you wish.
Part-1 Unboxing & Shipping List
We provide two versions kit and three main control boards option. Users can choose according to their needs and budget.
| Standard Version | Ultimate Version |
 |
 |
The product shipping list contains the following content.
 |
Robotic arm body + chassis |
 |
Suction cup |
 |
Nylon cable tie*6 |
 |
Depth camera |
 |
Depth camera bracket |
 |
Robotic arm expansion board |
 |
PCB patch antenna *2 |
 |
Patch antenna acrylic plate |
|
micro USB data cable |
 |
Upper elbow Type-C data cable |
 |
Wireless handle + AAA battery |
 |
30*30*30 wood block (red/green/blue/ yellow) |
 |
Power adapter + cable |
 |
30*30 block (red/green/blue/ yellow) |
 |
30*30*60 block (red/green/blue/ yellow) |
 |
DOFBOT Pro map |
 |
40*40*40 block *1 |
 |
Screwdriver |
 |
3D calibration map |
 |
Screw/copper column/wire package |
 |
Mechanical code & checkerboard card |
 |
OLED |
If you choose the Ultimate version, your package will also include the following content.
If you choose a different Jetson development board, your package will include the following content.
Part-2 Assembly
We provide customers with assembly videos and instruction manuals, so users can complete the assembly quickly.
Part-3 Quickly Started | Experience DOFBOT PRO
We provide customers with assembly videos and instruction manuals, so users can complete the assembly quickly.
Part-4 Product Hardware Configuration
We provide customers with assembly videos and instruction manuals, so users can complete the assembly quickly.
We provide customers with assembly videos and instruction manuals, so users can complete the assembly quickly.We provide customers with assembly videos and instruction manuals, so users can complete the assembly quickly.We provide customers with assembly videos and instruction manuals, so users can complete the assembly quickly.We provide customers with assembly videos and instruction manuals, so users can complete the assembly quickly.
DaBai DCW2 Depth Camera
The camera parameters are as follows.
|
Parameters |
Specifications |
|
Working distance |
Normal energy level mode: 0.15-3m High energy level mode: 0.15-5m |
|
Device size |
89.82mm*25.1mm*25.1mm |
|
Communication/power supply method |
Type-C/USB 2.0 |
|
Relative accuracy |
<1%@1m; <1.3%@2m |
|
Depth resolution @frame rate |
640*400@5/10/15fps; |
|
Depth FOV |
H91°V62°± 3° |
|
Resolution @ frame rate |
16:9 1920x1080@5/10/15/30fps&MJPG 1920x1080@5fps&YUY2 |
|
Color FOV |
16:9: H86° V55° D93.5°+3° |
|
lmage format |
MJPG/YUY2 |
|
Applicable scenarios |
Indoor/outdoor |
|
Safety |
Class 1 laser |
|
Camera principle |
Binocular structured light |
|
Weight |
88.1g |
|
Certification |
CE/FCC/KC/RoHS/Class |
High quality bus servo
The robotic arm body is equipped with 5PCS 15KG bus servos, and gripper adopt 1PCS 6KG bus servo. The servo can read position status information.
Built-in metal gears, wear-resistant and durable, anti-reverse connection interface can be used for cascade control.
|
Model |
15KG bus servo (Body servo) | 6KG bus servo (Gripper servo) |
|
Size |
44.37*23.06*35.12mm | 40.6*20*41.8mm |
|
Working voltage |
6.0~7.4V DC | 4.8~6.0V DC |
|
Rated torque |
>=15kgf.cm at 7.4V | >=6kgf.cm at 6V |
|
Rotation range |
300°±15° | 180°±10° |
|
No-load current |
=<310 mA at 7.4V | =<120 mA at 7.4V |
|
Stuck current |
=<3.2 A at 7.4V | =<1.7 A at 7.4V |
|
Servo accuracy |
=<1° | =<1° |
|
Control mode |
UART serial port command | UART serial port command |
|
Communication Baud rate |
115200 | 115200 |
|
Servo ID |
1.2.3.4.5 | 6 |
|
Gear type |
Metal gear | Plastic gear |
|
Interface model |
PH2.0-3PIN | PH2.0-3PIN |
10.1-inch Touch Screen Display
This 10.1-inch display adopt wrapped metal case, hide the connection, overall appearance is clean and tidy, and it supports pitch angle adjustment.
Note: Jutst for ultimate version
| Material | Metal cold-rolled plate | Process | White fine sand paint |
| Design | Case wrap design | Screen pitch angle adjustment | 0-90°(Actual adjustable angle depends on the installation situation) |
| Wire interface | Hidden design | Screen size | 239.4*157.4*11.3 +0.2mm(Excluding case bracket) |
| Screen size | 10.1-inch screen | Screen type | IPS screen |
|
Backlight adjustment |
Key switch adjustment | Screen resolution |
Maximum resolution: 1920X1080 Recommended to use 1280X800 |
| Touch screen type | Capacitive touch screen(integrated SlS9200) | Power supply | Micro USB(5V) |
| Total power | 5.29W (100% brightness) | Video input interface | HDMI (Up to 1080P) |
| Overall size | 247.1*160.4 29.9mm | Overall weight | 1009g |
Part-5 About development system
DOFBOT-PRO is developed using ROS robot operating system, which is an open source operating system for robots. lt provides the services that an operating system should have, including hardware abstraction, implementation of common functions for underlying device control, inter-process messaging, and package management. lt also provides tools and library functions for acquiring, compiling, writing, and running code across computers.
The system versions used by the three development boards are different.
|
Jetson Nano B01 |
Ubuntu18.04 |
ROS1_Melodic |
|
Jetson Orin Nano 4GB |
Ubuntu20.04 |
ROS1_Noetic |
|
Jetson Orin NX 8GB |
Ubuntu20.04 |
ROS1_Noetic |
Since the computing capabilities of the three development boards are different, some functions of DOFBOT-PRO are different when using different development boards.
As shonw below.
|
Version |
Jetson NANO B01 4GB Version |
Jetson Orin NANO 4GB Version |
Jetson Orin NX 8GB Version |
|||
|
Program start time |
Program running frame rate |
Program start time |
Program running frame rate |
Program start time |
Program running frame rate |
|
|
Robot startup |
62S |
|
57S |
|
48S |
|
|
2D-Face tracking |
4S |
10FPS |
5S |
13FPS |
5S |
30FPS |
|
2D-Gesture recognition to grab block |
7S |
6FPS |
6S |
22FPS |
5S |
30FPS |
|
2D-Fingertip trajectory recognition |
10S |
5FPS |
8S |
20FPS |
4S |
30FPS |
|
MoveIt |
45S |
6FPS |
35S |
31FPS |
16S |
31FPS |
|
3D-Yolov5 garbage recognition and sorting |
64S |
5FPS |
26S |
10FPS |
11S |
12FPS |
|
3D-Mediapipe gesture machine code distance sorting |
9S |
6FPS |
6S |
10FPS |
3S |
15FPS |
Part-6 Important Functions
Adopt wrapped metal case, hide the connection, overall appearance is clean and tidy, and it supports pitch angle adjustment.
Depth Vision Arm-camera integration
The end is equipped with a high-performance binocular structured light depth camera, which enables DOFBOT-PRO to calculate the distance, shape, height volume and other information of the object in 3D space based on the RGB data, position coordinates and depth information of the target object, thereby improving the autonomy and intelligence of the robotic arm to achieve more Al projects.
3D depth point cloud recognition
DOFBOT-PRO can obtain the depth map, color map, and point cloud map of the detection environment through the corresponding APl of the depth camera. Then, obtain the RGB data, position coordinates, and depth information of the target object.
Depth camera distance measurement
By obtaining the depth point cloud data of the object, the distance between the object and the depth camera can be obtained. Then, the object can be located, sorted, and tracked in 3D space.
Target object height measurement
By obtaining the depth point cloud data of the object, the height of the object can be identified. Then, objects with abnormal height can be removed.
Regional target volume measurement
By obtaining the depth point cloud data of the object, the shape and height of the object can be identified. Then, the volume of the object can be measured.
Sorting And Gripping In 3D Space
Sorting and gripping in 3D space [machine code/shape/color]
By calculating the 3D coordinates of the machine code in the image, the robotic arm can sort it to the specified location according to the machine code content. (The shape and color sorting process principles are the same.)
Garbage sorting
Yahboom provides blocks with garbage patterns. By loading the corresponding model, DOFBOT-PRO can quickly identify different "garbage" and place them in the corresponding classification area.
Tracking And Gripping In 3D Space
Tracking and grasping in 3D Space
Hold the block and move it in front of the robotic arm, robotic arm will track the block. When the block stops, the robotic arm will grab it and place it in the corresponding position.
KCF tracking and grasping
Through the KCF algorithm, the robotic arm can track the specified target object. When the target is stationary, the robotic arm will grab the target and place it at the corresponding position.
Movelt Simulation Control
DOFBOT-PRO supports Movet simulation, which can control the robotic arm and verify the algorithm in a virtual environment. Effectively reducing the requirements for the experimental environment and improving experimental efficiency.
URDF kinematics simulation model
Simulate according to the model of the real object in proportion, use the Movelt tool for visual control. Can realize dragging, preset position control, robotic arm movement obstacle avoidance and other functions.
Movelt simulation control/trajectory planning
When controlling the robotic arm movement in the simulation interface, the real robotic arm can make corresponding movements.
Color block sorting
Place the block in the map recognition area, robotic arm will automatically distinguish the current color and grip the block to the corresponding color area on the map.
Face recognition and tracking
Detect whether there is a face in the current picture, mark and follow it after recognizing the face.
Part-7 Multiple remote control methods
DOFBOT-PRO support APP, handle and PC web control.
Summary
Dofbot Pro is a 6DOF 3D vision robotic arm, equipped with a binocular structured light 3D depth camera. The unique structural design integrates the robotic arm with 3D vision. It can accurately identify the shape, distance, height, and even volume of objects. Supports calculating the position of objects in three-dimensional space, achieve accurate positioning, grasping and tracking objects.
Compatible with Jetson NANO B01, Jetson Orin NANO 4GB, Jetson Orin NX 8GB board. Through the GPU of the Jetson board for machine learning model inference, DOFBOT PRO can realize garbage/color/object recognition, etc. Support adding 10.1-inch touch screen to facilitate user development and debugging ROS robots. By integrating the ROS robot operating system and inverse kinematics algorithm, simplifies the complex motion control of 6DOF robotic arm, and can realize robot coordinate control, motion planning, robot collision detection, gripping and sorting, etc.
Overall, DOFBOT Pro is a flexible platform for ROS education and research.