A High School Project: Computer Vision and Robotics to Enhance Waste Recycling

Monday 8/7/23 04:05am
|
Posted By Rajan Mistry
  • Up0
  • Down0

Snapdragon and Qualcomm branded products are products of
Qualcomm Technologies, Inc. and/or its subsidiaries.

Waste recycling is still inefficient in many countries. In particular, even in 2023,many valuable materials like precious metals are still melted down or compacted and buried. Unfortunately, the cost of identifying and extracting these metals on a recycling facilities’ conveyor belt is still too high, and not automated enough, to make it worth the effort for the recycling industry.

Two 10th graders from Del Norte High School, CA, decided to take the initiative to tackle this problem. Along the way they learned some cool technology - computer vision and robotics automation.

Meet Rachit Jaiswal and Ethan Tran, who founded the “Tech Space Club” at Del Norte High School. These students convinced local industry experts, like Qualcomm Technologies to advise, support and accelerate their journey into learning, implementing, and testing technologies to solve this practical problem. The Tech Space Club is now engaging with many industrial experts in various domains to help design and build their project.

The project presented by Rachit and Ethan is a robotic arm with an embedded vision camera. It also includes an AI application to learn, detect, and pick out precious metals in collected garbage. The concept was built using the Qualcomm Innovators Development Kit (also named QIDK). QIDK offers several code samples, tutorial videos, pre-trained models, and sample applications for students to quickly and conveniently learn about and take advantage of the Qualcomm AI Stack for their projects.

During our conversation, Rachit, President of the Tech Space Club, said they chose QIDK because "it had all the prerequisites needed for our project while allowing us to test on a device that will function in the same way as our end product." The students were able to test multiple computer vision models like MobileNetSSD, EnlightenGAN, and YoloNAS for their prototype and quickly went from idea to prototype using QIDK.

Although out of scope for their current project, the students spent some time porting MobileBERT and GPT2.5 models on QIDK to explore how a smart device can execute natural language processing inference on the edge.

For their next steps, the students plan to contact nearby waste management facilities to test their prototype in a production environment. The Tech Space Club also plans to build additional projects using QIDK.

We bet that as more AI projects are initiated by the students, QIDK will continue to be a valuable resource for the success of the club. We look forward to welcoming new student projects like this one, which illustrate how Qualcomm Technologies envisions the rise of new technologies by empowering a new generation of developers.

Please join us in encouraging Rachit and Ethan in their journey to learn about and practice the concepts of edge AI, computer vision, and robotics in their Tech Space Club adventure.

To learn more about QIDK and object detection code samples, please visit:



Qualcomm branded products are products of Qualcomm Technologies, Inc. and/or its subsidiaries.