The artificial intelligence suit is a suite based on visual recognition + collaborative robotic arm
myCobot320 Artificial Intelligence Suite 2023 Edition
myCobot320 six-axis collaborative robotic arm
The biggest difference between the artificial intelligence suit myCobot 320 version and the previous version is the new support for the myCobot 320 six-axis collaborative robot arm.
The myCobot series of manipulators is also a leading product of Elephant Robot , among which the 320 model is a six-axis collaborative manipulator developed for users' independent programming . The myCobot 320 robotic arm itself weighs only 3kg, but its payload has reached 1kg, and its working radius has reached 350mm. It is small in size but powerful in function.
Open ROS simulation development environment; built-in perfect kinematics forward and reverse algorithm module; free matching visual development application;
Intuitive structural analysis of mechanical motion principles; enterprise-level communication protocol specification and application; 12 standard 24V industrial I/O interfaces;
Support the mainstream control language interface in the industry; develop the communication protocol of the manipulator; rich terminal expansion accessories;
The three major advantages of the myCobot 320 robotic arm in terms of ease of use, safety, and economy make the myCobot 320 robotic arm a cost-effective choice for commercial applications, educational research, and creative development.
suit new upgrade
320 artificial intelligence suit 2023 version, 5 major visual recognition algorithms, 7 major application scenarios, 2 grasping methods, super large load 1KG, M5 and PI 2 kinds of adaptive robotic arms, 8 major learning points and support visualization software, it is a positioning grasping An entry-level artificial intelligence package that integrates the modules of picking, automatic sorting, and item grabbing. Based on the python platform, the control of the robotic arm can be realized through the development of software. It is easy to learn, and can quickly learn the basics of artificial intelligence, inspire innovative thinking, and understand the open source creative culture.
Compared with the original artificial intelligence suit, it has significant improvements mainly in the following aspects
Larger working radius and load
Thanks to the support for the myCobot 320 six-axis collaborative robot arm, the artificial intelligence suit myCobot 320 version has broken through the limitations of the working radius and load of the previous artificial intelligence suits. The working radius has come to 350mm from the previous 280mm, and the load has also changed from the original 200g. Increased to 1000g. Larger working radius and load also represent richer application scenarios.
Gripper item identification, intelligent grabbing
Different from the artificial intelligence suit that only supports the grabbing method of the suction pump, the artificial intelligence suit myCobot 320 version adds support for grippers.
The high-performance gripper and the high-load myCobot 320 enable the artificial intelligence suit myCobot 320 version to perform object identification and intelligent grasping, and identify and grip a variety of objects of different sizes and weights.
5 major visual recognition algorithms
The artificial intelligence suite is based on robot vision and hand-eye calibration, and supports the following five major visual recognition algorithms
color recognition
Color recognition is realized through computer vision technology, mainly by analyzing and processing the RGB values of pixels (combined values of red, green and blue colors) for color recognition. Common color spaces include RGB, HSV, YUV, etc. Among them, the RGB color space is the most commonly used color space. The method based on the color space mainly judges the color category of the pixel by setting the threshold of the color space.
shape recognition
Shape recognition refers to the automatic recognition of objects of different shapes by analyzing and processing the shape and structure of objects in the image in the field of computer vision. Extract the outline of the object through image processing technology, and perform feature extraction and description, such as boundary points, curvature, perimeter, area, etc., and then match it with a predefined template or known shape to achieve shape recognition.
QR code recognition
Aruco QR code recognition is a computer vision technology based on OpenCV for fast and accurate recognition and positioning of QR codes. Use the functions in the Aruco library to process and analyze the input image, extract the Aruco QR code in the image, and recognize the code and location information of each QR code. In the process of detection and recognition, the Aruco library performs operations such as binarization, corner detection, and codeword extraction on the input image to improve recognition accuracy and speed.
feature point recognition
Feature point recognition is a key technology in computer vision, which is used to find points with unique properties and repeatability in images, and use them as the basis for tasks such as image matching, object tracking, and 3D reconstruction. For selected feature points, extraction and description are performed in the image. Commonly used feature description algorithms include SIFT, SU RF , ORB, BRIEF, etc. These algorithms are able to extract feature points with unique properties from images and represent them as stable and comparable feature descriptors.
YOLOV5 recognition
YOLOV5 is a deep learning -based target detection algorithm that uses a detection method called "one-stage", which can simultaneously predict the location and category of multiple targets in an image. YOLOV5 uses a ResNet-like backbone network to extract image features for easy target recognition. Different from traditional two-stage detection methods such as Faster R-CNN, YOLOV5 combines feature extraction and target detection, which greatly improves the detection speed.
visualization software
In addition to the above hardware functions, the artificial intelligence of the Elephant Robot has also made a lot of effort in software control, supports visual software operation, can choose the corresponding visual algorithm, fully automatic recognition, grasping and placement, easy to watch the effect display, and supports step-by-step operation, It can only identify, only grab, and only place, so that users can understand the principle of each process.
With its highly open and visualized UI interface, the artificial intelligence suite is more handy whether it is used for the development and verification of scientific research and education, or the decomposition and learning of beginners, or the creative application of makers.
Parameters and comparison of old and new
| parameter | |
| name | 320 Artificial Intelligence Suite 2023 Edition |
| total measurement | 521mm*637.5mm*600mm |
| recognition speed | Color/shape/QR code: 300ms Feature point/yolov5: 600ms |
| Recognition accuracy | 3 mm |
| camera name | USB distortion free camera |
| image pixels | 2M1080p |
| Supported image formats | MJPG/YUY2 |
| pixel size | 3.0umx3.0um |
| Maximum frame rate | MJPG：1920 1080@30fps YUV：1920 1080@30fps |
| End fittings | Adaptive Gripper Pro, myCobotPro Single-head Suction Pump |
| load | 1kg |
| USB protocol | USB2.0HS/FS |
| Supported Resolutions | 1280X720,640X480,320X240 |
| power supply | DC5V90mA |
| Field of view | 110° no distortion |
| supported system | Windows7/8/10 LINUX (includeuvc)/ Raspberry Pi |
| Compared | ||
| parameter | artificial intelligence suit | 320 artificial intelligence suit |
| Robot arm reach | 280mm | 350mm |
| Arm load | 200g | 1000g |
| item identification | not support | support |
| Smart gripping | not support | support |
| Jaws | not support | support |
| suction pump | Support<=200g | Support=1000g |