One of the things that we realized was that we can’t actually make use of four mics with the I2S connections. This is due to our original idea banked on being able to put in two inputs per I2S port but the problem is that the inputs from the mics only get read as a single mixed input which we can’t separate out. That means that we are forced to use only one mic per input which can potentially hurt the accuracy of acoustically locating the speaker. We still feel confident that we can still make a good approximation with only the two mics. We are also considering, as an alternative, using the three I2C ports (one of the Nano’s four I2C ports is already being used for the motors) for audio input, so that we can employ three mics. However, this would mean having to use analog mics and analog-to-digital converters as opposed to the digital mics we’re currently using, so we’re going to see if the two-mic setup is adequate before making this change.

On a happier note, we’re also almost done with the mechanical prototype. We’re still waiting on a couple of things like longer MIPI CSI-2 cables so the camera can extend all the way, and some micro servos so the module that moves up and down is smaller. But the elevator shaft and rotating base is prototyped.

For the software component, the biggest thing that we have to start thinking about is the integration between different components. We have a decent start to each of the individual components so thinking about different concurrency models that we can use is the next big step. We have the most basic version which is to use threading but there is also the option of using a golang library for python that allows us to make use of the concurrency model that exists in go.