Status Report #9
Joel :
Once we got the door to camus our next problems was fixing the issue that came when transporting. We broke some of the frame off and weren’t able to properly stand the door up. So we took it to makerspace, where we found a bunch of spare wood and tools to fix the frame. We measured the frame, and what we needed and designed something we thought would work on keeping the door up. We executed our plan well and had fun doing it. As ECE majors we don’t really work too much with our hands other than building circuits and doing things like that. But this gave us the opportunity to cut wood and drill things down and be even more creative. I actually learned how to use something called an oscillating saw, which makes straight cuts into different materials by oscillating a saw. After only a day, our door fell down, and we had to come up with a new plan for fixing it up. We upgraded the frame that we already had, and made it more stable than it already was. After this we cut into the door, finding space to put all our materials in, such as the Pi, and circuit. We kept implementing and re-implementing to make sure everything was perfect.
Omar:
As we were approaching the final weeks, we started thinking about integrating all our parts together. We still needed to do some work with the pipeline between the different subsystems, though, as the connection between the pi and all the sensors wasn’t working at first. However, with some time debugging, we were ready to go with the raspberry pi. I did some work with the web application, as we needed to update our design to the new webapp design I created and also integrate the texting validation feature into the door system. This week was filled with working on a bunch of different systems and debugging when connecting them didn’t work. Integrating was a slightly slow process, but it was cool seeing everything come together.
Chinedu:
In this week, the pipeline had to be updated in a major fashion. We started to the process of integrating this week and confirmed that this indeed was the toughest part of our Capstone project. In order to asynchronously poll the camera for frames to send the ML suite, two process had to access the camera at a given time. The streaming library we intended to use UV4L, did not allow for another process to access the RasPi camera while the stream was being sent to the client. For this reason, I had to completely change streaming and camera polling on both the front and back ends of the system. I used the PiCamera interface to create a video feed and allow for the stream to “screenshotted” at any given time by the camera poll to be sent to the ML suite. In addition, I created a steam handler which constantly sent video frames to the front end using the Flask app.