The project we created is a navigation assistant for the visually impaired. It uses an SSDLite recognition model, running on a Raspberry Pi computer chip. We have trained this model on recognizing crosswalk lights to help the visually impaired in daily navigation – specifically crossing the street. In its current form, it lies in a self-designed 3-D printed case.. The case has accommodation for a portable power source, allowing it to be taken around outside. To build it we used a Raspberry Pi chip, camera, and hat; a basic electronics kit; a soldering iron; and a 3-D printer. We used tensorflow as our basis for the machine learning aspect. Through google colab, we trained multiple models on various self-collected datasets to determine which model was optimal for real-time analysis. In the end we chose SSDLite model, a single shot detector algorithm, as a balance between speed and accuracy. As we are still in the prototyping phase, it can only detect crosswalk lights, but as we improve it, we will train it on different objects to provide a better-rounded navigation assistant. What it does, is it receives a video input from the Pi camera, which the image recognition model processes, and identifies crosswalk lights. It then, using vibration motors informs the user of whether the light is green or not, and in which direction it lies. We believe this innovation is important, as there are few projects like it, and there is always room to improve aid for the less fortunate.

What inspired you (or your team)?

What initially inspired us was a youtube video in which a visually impaired radio presenter, film critic, and youtuber by the name of Tommy Edison filmed himself crossing the street. In the video, it was apparent that he was struggling, especially with direction, as it was a big intersection. At one point, he almost walked into a post on a divider in the road. What inspired us further is when we visited the Canadian National Institute for the Blind community office. While there, we talked to a woman by the name of Christine Malec, who has been blind her whole life about our project. She said that it was often difficult for her to navigate large intersections, and ones she was not familiar with. That conversation very much inspired us to continue and develop the project, so we can make a difference, and fill a void that is very obviously there.