Manuals+

DOFBOT AI Vision Robotic Arm

Artificial intelligence, Camera and robot arm 2 in 1

About DOFBOT

DOFBOT is an AI vision robotic arm with Jetson nano as the main control, Open Source CV as the image processing library, mainstream Jupyter Lab as the development tool, and Python3 as the main programming language. The camera is integrated with the robotic arm, and the "2 in 1" design makes the visual recognition of the robotic arm more flexible. It can not only realize color recognition, tracking and grabbing, but also recognize and interact with human body features, and even train garbage classification models for garbage sorting. Through the ROS robot control system, the complex motion control of the 6-DOF serial bus steering gear is simplified. The official provides a guided quick-start tutorial to help you easily control the DOFBOT AI vision robot arm.

Based on JETSON NANO'S powerful AI computing capabilities

NVIDIA Jetson Nano is a small but powerful computer that can run multiple neural networks, object detection, segmentation and speech processing applications at the same time. Equipped with quad-core cortex-A57 processor, 128-core Mazwell GPU and 4GB LPDDR memory, Jetson Nano has sufficient AI computing power, provides 472 GFLOP computing power, and supports a series of popular AI frameworks and algorithms, such as TensorFlow, Pytorch, caffe/caffe2, Keras, MXNET, etc.

JETSON NANO 4GB (optional)

Artificial Intelligence Vision Robotic Arm

The first-person perspective controls the robotic arm, making visual recognition more flexible

01 - Artificial intelligence / Camera and robot arm 2 in 1

Camera on the robot arm: 3D simulation modeling, complex and powerful algorithm FPV vision, stronger experience.

Separate camera: Convert 3D scene to 2D plane. Simple algorithm, suitable for getting started.

02 - FPV HD video control

Remote control interface displaying angle adjustments and stop functionality.

Cross-platform interconnection control

Multi-function mobile phone APP (iOS/Android)

A variety of AI gameplays are built-in, and the robot arm can be watched in real time and supports at the same time: gesture interaction, tracking game, garbage sorting, etc.

Available APP functions include: Gesture Interaction, Color Recognition, Tracking Game, DOFBOT Setting, Tracking Game, Gesture Interaction, Color Recognition, Color Calibration, Advanced Settings (Beta), Servo Calibration, Action Groups.

PC host computer control

In addition to the video screen of the FPV camera, the host computer has also added a 3D simulation model of the robotic arm, which rotates synchronously with the entity, allowing the robotic arm control theory and practice to be combined.

USB handle remote control

In addition to mobile phone APP and computer control, it also comes standard with a USB handle to experience the fun of combining multiple degrees of freedom and multiple keys.

Color recognition interaction

1) Color recognition tracking

In addition to standard colors, custom colors can also be extracted for color recognition and tracking of the robotic arm, so that the camera can follow the selected color automatically.

2) Catch Game

In addition to standard colors, custom colors can also be extracted for color recognition and tracking of the robotic arm, so that the camera can follow the selected color automatically.

3) Color capture

Place the wooden block in front of the camera, and the robotic arm will recognize the color of the wooden block and clamp the wooden block in the corresponding color area to the middle area.

4) Color interaction

Choose DOFBOT favorite color. It will follow the target color back and forth.

Model training / Garbage sorting

Basic function: The DOFBOT kit comes standard with multiple wooden blocks with trash patterns. By loading the garbage classification model, the camera can identify the type of garbage on the wood block and return the garbage name and classification data to the APP. The robotic arm sorts the "garbage" to the corresponding classification area on the map.

Developer function: You can also identify the area by calibrating the map, place multiple "garbage" blocks in the identification area for real-time recognition, and continuously sort multiple "garbage" to the corresponding classification area on the map.

Garbage categories: Other garbage, Recyclable, Harmful waste, Kitchen waste.

Human feature recognition and interaction

DOFBOT can recognize human features such as gestures and faces through the camera screen, and complete interactive actions such as grabbing, recognition and tracking.

1/Gesture interaction

DOFBOT can recognize multiple gestures and perform corresponding actions for interaction.

2/Gesture grab

Recognize digital gestures, stack the corresponding layers, and finally push down the stacked wooden blocks under the "fist" gesture.

3/Face recognition and tracking

Detect whether there is a face in the current picture, mark the face after it is recognized and follow the movement.

Robot motion control

01/Robotic arm custom learning action group

Due to the intelligent serial bus servo, DOFBOT can read and record the angle of each rotation after entering the action learning mode to realize learning and repeat the action group.

Fun fixed action group/02

There are 8 fixed action groups provided in the APP, you can click the serial number to preview, and click Run to start execution.

03/Synchronous teaching of robotic arm

This gameplay requires two DOFBOT robots for teaching control. Read the joint angle of the servo on the DOFBOT Master Machine side, and transmit it to the Slave Machine side in real time, allowing the Slave Machine to rotate synchronously according to the posture of the Master Machine.

6 degree of freedom inverse kinematics control/04

Decompose the servo motion control of the robot arm with 6 degrees of freedom, calculate the theoretical motion angle of each servo by inputting the target coordinates, and simultaneously control the motion of each DOFBOT servo in combination with the servo control protocol.

05/Operating system

The ROS robot operating system is a collection of tools, software libraries and protocols, designed to simplify the DOFBOT robot platform and build complex and powerful robots.

Powerful hardware configuration

Function on expansion board

  1. Micro USB interface
  2. I2C interface
  3. Cooling fan interface
  4. Button: K1+K2+RESET button
  5. Status Indicator
  6. RGB light
  7. Ultrasonic interface
  8. Serial port function selection
  9. PWM servo interface
  10. buzzer
  11. Bus servo interface
  12. PS2 handle receiver base
  13. T-type power supply interface
  14. Switch
  15. Serial port

Guided quick start

Guided teaching, follow the APP in three steps:

  1. Scan and configure network: DOFBOT connect to WIFI and pair by scanning the QR code.
  2. Servo calibration: By reading the angle of the steering gear, angle calibration of the robotic arm entity.
  3. Color calibration: For beginners and developers, we provide basic color calibration and professional color calibration functions to avoid environmental factors affecting the robotic arm.

Multiple remote control methods

APP remote control: iOS/Android APP

Provides various control interfaces for the DOFBOT robot arm.

PC host computer

Allows for programming and control via a host computer.

Jupyter Lab webpage online programming control

Enables online programming and control through a web interface.

ROS system

Utilizes the ROS operating system for advanced robotic control.

PS2 handle control

Supports control using a PS2 game handle.

Massive learning materials

Access extensive learning resources at: http://www.yahboom.net/study/Dofbot-Jetson_nano

Learning materials include:

Specific tutorials available:

Product specifications

Product Size

Dimensions: 303mm (length) x 135mm (width) x 473mm (height)

Product parameter

Microprocessor: Quad-core ARM A57+128-core NVIDIA Maxwell
Al computing power: 472GFLOPs
Operating system: Ubuntu 18.04 LTS + ROS_Melodic
Programming language: Python
Steering gear scheme: 15*6kg*5+6kg*1 Serial Bus Smart Servo
Input: Wide-angle camera / Emergency stop button / Distribution network button / Reset button / 2*IIC interface / 5V fan interface
Output: RGB indicator / OLED display buzzer / 6 bus servo interface / 6-channel PWM servo interface
Power scheme: 12V5A power adapter
Power connector: T plug interface
Remote control method: Mobile APP/PC upper computer/PS2 handle(PC port)
Way of communication: WIFI network
Safety protection: Overcurrent protection, Reverse connection protection
Mechanical arm material: Anodized aluminum
Assembled size: 303*135*473mm
Weight after assembly: 1256g
Degree of Freedom of Robot Arm (DOF): 5 degrees of freedom + gripper
Payload: 200g (Straightenable weight)
Load: 500g (Clamping handling weight)
Arm span: 350mm
Diameter of grabbed object: 1-6CM
Effective crawl range: Radius <=30cm, the area with the central axis as a semicircle
Repeatability: ±0.5mm
Voltage: 100~240V 50/60 HZ
Camera: 300,000 pixel 110 degree wide-angle camera
Resolution: 480p (640*680)
Frame rate: 30fps
Focus method: Focal length can be adjusted manually

Package List

Features

Reasonable and convenient first trial

All aluminum alloy bracket with 2mm thickness

Top hardware Configuration

Fantastic AI function

Models: 6 DOF, DOFBOT, DOFBOT AI Vision Robotic Arm, AI Vision Robotic Arm, Robotic Arm, Arm

File Info : application/pdf, 15 Pages, 1.70MB

PDF preview unavailable. Download the PDF instead.

2f0b3c55 dofbot

References

WPS 文字