I’m back in the Netherlands, right here in the heart of Delft, to update you on the progress we’ve made over the past few months. My holiday took me across Europe, but now I’m back at the Plastic Scanner headquarters, and let me tell you, exciting things are happening.
First and foremost, let’s talk about our prototypes. In our previous updates, you saw us working on the development boards 2.2. We made strides, experimenting with different components, particularly the ADC (Analog-to-Digital Converter). While the newer ADC we tested might be slightly less precise than the original Texas Instruments ones it’s readily available and fits perfectly into our open-source project.
Now, let’s fast forward to the present. We’ve transitioned to the development board 2.3, which retains the same footprint as its predecessor. The beauty of this version is its modularity. We’ve designed it with slits down the middle, allowing it to split into two parts: the sensor side and the processing side. This innovation means it can work independently with various processors, like Arduino Uno and even the ESP32.
To facilitate communication with different platforms, we introduced an I2C connector. This connector links to the sensor board, enabling seamless interaction with devices like the ESP32. This integration led us to create a compact housing unit! Picture this: a device housing the development boards, connected to an ESP32 board, with LEDs controlled via a separate LED board. This setup offers versatility and ease of use, allowing us to switch between prototypes effortlessly.
But that’s not all. We found that by using a backing material made of sintered PTFE on the scanner’s backside, we enhanced our accuracy significantly. This material reflects infrared light back through the plastic, providing a clearer signal. This advancement has opened new doors, particularly in applications involving laser-cut plastics. Clean, laser-cut plastics like PMMA, PC, PET, PS, and PP can now be identified using our scanner.
The most exciting news, however, lies in our venture into machine learning. We’ve taken our Plastic Scanner project to the next level by implementing TensorFlow Lite on an embedded device. This addition allows the scanner to interpret the type of plastic it’s scanning, providing valuable insights. While this feature is in its early stages, it marks a significant leap forward, hinting at the incredible potential our project holds.
It’s important to note that while we’ve made progress, our Plastic Scanner is not yet suitable for widespread use, especially in landfills. There’s still work to be done to enhance accuracy and broaden our scanner’s capabilities to encompass a wider range of plastic types.
If you’re as excited about this journey as we are, we invite you to join our Discord group. Additionally, you can explore all our project files on docs.plasticscanner.com or our GitHub repository. We’ve also crafted a TensorFlow Lite model, complete with Google Colab instructions, to showcase our commitment to transparency and collaboration.
As I delve back into our work, I’m excited for the future. Together, we’re shaping a world where plastic recycling becomes not just a dream but a tangible reality. Thank you for your support, and I look forward to sharing more updates in the coming months.
Stay tuned for the next chapter in our Plastic Scanner adventure. Until next time!