After a great initial design, and planning, the team realised that an initial test set-up was needed. Although we had a final design in mind, a prototype testing apparatus would allow us to set-up the sensors and get the electrical and software components fully functional before the final installation. The testing set-up would just need to mimic the enviroment that would be provided by the final design.

From a mechanical perspective, this initial protype would also show the designer how the rest of the team used and interacted with all of the various sub components, from LEDs to the raspberrry PI. The prototype would show areas that need to be improved upon as well as help establish the priority in terms of sub-component importance.

The images showcase just how quick and dirty this initial prototype was. The idea around the test set-up was to enclose the IR sensor in a dark setting. This would mimic the enviroment in the final enclosure. The IR sensor could then be adjusted and stabilized rather then waiting till the final enclosure to make all the adjustments.

All the resources were bought at Home Depot. Tape was used to limit the amount of light being let into the IR sensor. From this initial set-up, the team was able to get tangible results and begin studying the data from the IR sensor application.

A similar method was used for the Camera enclosure.

In terms of mechanical design, the project took some what of a pivot. Since the sensors and the detection part was not complete, this limited the scope and mechanism of the mechanical component of the project. Rather than having a automated removal of the clothing, instead two different concepts where looked at.

1. A easy to transport mechanism.

• Easily moveable

• Base created from wood
  • Manufacturability
  • Easy Sourcing
  • Versatile in order to accommodate sensor orientation
  • Use of metal will be overkill in terms of design as components being housed are relatively light
• Cone like structure
  • Allow the creations of ideal lighting for detection 
  • Shielded from external environment
  • Will house photovoltaic sensor as well NIR/VIS sensor

2. A box structure

The box structure adheres to the original design without the addition, of the mechanism to automatically remove the clothing. The design thrives on it’s manufacturability, as it will be relatively easy to build and/or machine. Using wood will also allow the team to really accommodate the sensor locations. A cover, (not shown) will allow the team to control the lighting and protect from the external environment.

Overall, the team is currently working towards idea 1 with idea 2 as a back-up.

Attempt 1

We were able to find a dataset composed of close up images of different types of fabrics. We decided to train a deep learning network to classify the type of fabric given such an input. As a starting point, we decided we want to classify between 3 different types – cotton, polyester, and wool. This is because these materials have the most images in the dataset and so we would get better results compared to other fabrics when training a neural net. We created a 80-10-10 split for each of the three materials for training, validation, and test set. We also converted all the images to grayscale because we believe colour is not a strong or useful feature for determining material. A VGG-16 model architecture was used because in the paper discussing this dataset and results, a VGG-M model was used and this is a larger version of that architecture. Our testing set achieved an overall accuracy of 76%. Class-wise, this was 76% recall for cotton, 89% for polyester, and 39% for wool. Wool had considerably less images compared to the other classes and for our next iteration we will attempt to account for this class imbalance.

Today, we hooked up the spectral sensor to Arduino Uno through i2c connection and recorded the outputted spectral colour values for 3 different types of textiles (cotton, polyester, wool), while controlling the distance of the textile from the sensor, the lighting (natural), and recorded 3 trials for each.

The colour of the material has an effect on the spectral output, therefore we ensured to compare only like colours.

The next steps would be to create the setup in order to begin calibration.

Stay committed to your decisions, but stay flexible to your approach.

— Tony Robbins.

We came together and made a game plan for the project. The photodiode we had ordered last term is not giving a response – we suspect it is broken. A quick look at Digikey gave way to an unfortunate realization…the sensor is no longer stocked.

But, worry not. We are looking at another approach. Following a “flashing yellow light” warning, we have turned our attention towards finding a camera vision and image processing solution to distinguish between various textiles. This solution uses more of the expertise of the group, as well as seeming more plausible to complete in the next 5 weeks.

On the side, we will continue to investigate the sensor issue through oscilloscope testing as well as ordering an IR sensor breakout board.