We spent a lot of time together on Zoom this past week for integration and while Heather assembled the physical device. First and foremost, I worked with Edward to edit my audio detection code to be able to communicate with main. Before Heather added in the audio portion of the device, I wanted to make some optimizations. The acoustic location was updating the direction of the speaker far too often — all the readings would be around the same value but +/- a few degrees. This made the motors jitter back and forth with every new reading, so I modified the code to only update the angle if it detected a significant enough change in direction. Then, we noticed the acoustic location was occasionally returning an angle completely off from what it should have been reading (e.g. within a sequence of values around 190, it would return one or two values around 0). I added in some code to only update based on the median of a sequence of values being read in. For both of these changes, we had to play around with different implementations and values (e.g. using the mean as opposed to the median, reading in a sequence of length 20 as opposed to 10, how long to sleep in between readings in a sequence, etc.). Heather also helped with editing the code such that, by communicating with main, we would know not to update the angle of detection values when the motors were working, since they’re physically near the mics and produce some noise.

In the meantime, I got started on creating the final presentation slides. This is what we’ll be working on in the coming week, as well as the final paper.