Dalian Dayoushang Group launched an intelligent robot arm vision system solution based on Intel products

On November 3, 2021, Dalian General Holdings, a leading semiconductor component distributor dedicated to the Asia-Pacific market, announced that its subsidiary Youshang has launched a vision system solution based on Intel (Intel) NCS2 neural computing stick, which is suitable for intelligent robotic arms .

Dalian Dayoushang Group launched an intelligent robot arm vision system solution based on Intel products

Figure 1 – The Display board of Dalian Dayou Shang’s intelligent robotic arm vision system solution based on Intel products

A robotic arm is a machine that can flexibly replace human hands and arms. It is usually an automatic operation device that grabs, carries objects or operates tools according to a fixed program. With the continuous breakthroughs in machine vision, AI and other technologies, the development of intelligent robotic arms is also growing. Dalian Dayou is still based on the vision system solution of Intel NCS2 neural computing stick, which is applied to the robotic arm, which can accurately identify objects and make the process of grasping and handling more intelligent.

  

Figure 2-Scenario application diagram of Dalian Dayoushang’s intelligent robotic arm vision system solution based on Intel products

The core principle of this solution is to use D435 to identify the color image combined with the deep learning object localization algorithm YOLOv3 to find the object position (X, Y). When CPU computing power is insufficient, Intel GPU or VPU (NCS2) can be used to accelerate the calculation, and then the depth (Z) information of the corresponding position of the depth image can be recognized by D435, so that the robot arm can be correctly extended to the correct spatial position. (XYZ) Grip and grab the item.

Figure 3 – Block diagram of Dalian Dayou Shang’s intelligent robotic arm vision system based on Intel products

Intel NCS2 is a VPU (Vision Processing Unit) with a Neural Compute Engine that provides eight times the performance of the previous generation. Deep learning neural networks such as Caffe, TensorFlow or MXNet can be directly integrated with the OpenVINO toolkit on NCS2. And these machine learning frameworks are also optimized for new deep neural network (DNN) inference engines. With a PC and Intel NCS2, developers can jump-start the development of AI and computer vision applications.

NCS2 runs on USB 3.0 and requires no additional hardware, enabling users to seamlessly deploy models trained on their PCs to various edge devices without the need for internet or cloud connectivity. Whether developing smart cameras, drones with gesture recognition, industrial robots, or the next generation of smart home devices, NCS2 gives developers the capabilities they need for rapid prototyping.

Core technical advantages:

Intel RealSense D435 camera:

Low-cost 3D binocular depth camera;

Provide a complete SDK to quickly integrate with the system;

It can scan quickly and provide point cloud information;

Independent innovation functions can be developed through ROS integration;

Intelligent 3D object recognition.

Intel OpenVino Toolkit:

The trained model can be optimized;

Support training frameworks commonly used by the industry and academia;

Can be quickly deployed to Intel’s hardware platforms such as CPU, GPU, VPU, FPGA;

Provide common pre-training models such as SSD, YOLO, etc.;

Provide application examples of C++ and Python to shorten the program development cycle.

Intel Neural Compute Stick 2 (Intel® NCS2):

Build and scale with exceptional performance per watt per dollar on the Intel Movidius Myriad X visual processing unit (VPU);

Support rapid development on Windows 10, Ubuntu* or macOS*;

Develop on common frameworks and out-of-the-box sample applications;

Operate without connecting to the cloud;

Prototype on low-cost edge devices such as Raspberry Pi*3 and other ARM* host devices.

Program Specifications:

3D camera: Intel RealSense D435 Camera;

Operating system: Windows 10;

Intel NUC Rugged Board with Core i3/i5 Processor;

AI Inference Kit: Intel OpenVino Toolkit 2020.03;

Memory: 4GB or more;

Transmission interface: USB 3.0;

Neural Compute Stick: Intel Neural Compute Stick 2 (Intel NCS2).

The Links:   2MBI150NC-060-10 2MBI450U4E-120