Industrial AI and Robotics Controller System in Waste Recycle Industry
Project Introduction
Governments and industries across the world are raising the standard of recycling. Massive amounts of collected waste require precise material sorting, from small sized light objects such as bottle caps, to complex material mixtures like paper, plastics and metals. There are also essential sorting considerations for domestic hazardous materials. But, with accurate waste categorizations and separation, we can prolong the lifespans of incinerators, reuse valuable materials, and reduce greenhouse gas emissions. Moreover, billions of dollars worth of recyclable materials goes unrecovered annually. Therefore, how to mitigate the impact to the environment is an important issue for government and industry to focus on.
A recycling center customer was looking for a high-performance, industrial-grade Artificial Intelligence (AI) solution that could successfully function in harsh environments. The customer needed an automated solution to help identify certain materials on the recycling line.
System Requirements
A recycling center end user needed an AI solution to help identify and filter the various materials on its recycling line such as glass, plastic, cans, etc. Recyclable materials are brought to the center, sorted by machines, poured into bins, and then moved to final destinations for processing. In its search for an AI solution, the end user was already connected with an Advantech Channel Partner, thereby opening the door for an Advantech solution.
The customer needed help creating a new solution to enhance recycling performance by accelerating object identification and classification capabilities. To do this, they needed an industrial AI solution to take the place of their previous, less-effective commercial hardware solution. The new AI solution required high computing performance for processing on site. Advantech’s MIC-770 + MIC-75G20 solutions were the best option to meet the customer’s high-processing edge computing requirements.
The end user needed a high-performance processor system that was able to accommodate one GPU card, and two 4-port Power over Ethernet (PoE) cards to connect eight cameras. The hardware for the system also had to be able to successfully function in the industrial, harsh environments of a recycling plant.
System Description
Advantech’s MIC-770 and MIC-75G20 incorporate Intel’s 9th Gen. CPUs and NVIDIA’s high performance GPU cards. Together they provide one of the most compact-sized IPCs on the market, and perfectly meet edge video AI requirements. MIC-75G20 is a GPU expansion module and accommodates one high-performance GPU card and one 4-port PoE frame grabber card, connected with four PoE cameras. By adding additional 4-port PoE FIO in MIC-770, the eight cameras were able to capture a lot video data.
Huge amounts of images are able to be captured and processed by the GPU card and analyzed by the user’s AI model to identify and classify objects. Instructions are then delivered to the main controller via the MIC-770, which is a compact box computer that acts as an expansion slot module. Instructions precisely tell the robotic arms to pick-and-place specified items to a designated area. All actions can be done at eye-blinking speed, enabling the system to handle more than 80 items a minute.
Project Implementation
The individual components used were MIC-7700Q + MIC-75G20 with NVIDIA GPU card + PCIe-1674 + 4-port PoE advance FIO. End user requirements included a high-performance processor system that was also able to accommodate one GPU card and eight ports for Power over Ethernet (PoE) cameras. Advantech’s MIC-7000 series PCs met these requirements with high-performance processing, multiple I/O interfaces, and flexible expandability with the integration of i-Modules.
System Diagram
Featured highlights:
- High Performance system
- Harsh working environment with vibration
- Able to accommodate GPU and Plugin card
- Fast turnaround time
- Strong technical support
Conclusion
Advantech already had the needed hardware in its inventory and was able to work with our Channel Partner to prepare a complete solution bundle for the end user. Our inventory and Channel Partner bundle meant a quick turnaround time for the customer to get up and running. Advantech also prepared additional parts for the solution for faster repairs. By utilizing Advantech to create an automated AI solution to identify certain materials on the recycling line, the customer improved their entire process with more efficiency and accuracy. Additionally, with industrial-rated products, the end user did not need to worry about the solution functioning properly in harsh industrial environments.