The bulk of this week was spent working on our design review, as we realized just how significant of an undertaking it is and realized that it would be in our significant interests to start working on it before we got too deep into the project. This ultimately proved beneficial, as some significant design changes have been proposed that we will need to finalize before finishing the report this week. The biggest of these changes is with regards to the communication protocol between the boards. We realized that I2C may be susceptible to too much noise at the distances we need it to run across, and we also realized that the Micro only has one I2C port that is already being used to collect the sensor data. We believe that we can tackle this challenge, but we are looking into alternatives like SPI and TTL over UART to make sure that we have backup plans.
The design review occupied most of our time this week, but we were still able to make some progress on the project as a whole. We have managed to collect data from the gyroscope, but we still are working on processing it into a usable form. We have managed to perform basic object detection with the Jetson using the pre-trained SSD model from the Jetson-Inference library, and we have begun working on our own implementation. We are currently researching whether it would be more reasonable to retrain an existing model with new images than to create one from scratch. We also expanded our MIDI boilerplate code and managed to get analog inputs to drive numeric MIDI signals as well.
As of now we are slightly behind schedule. We did not anticipate the design review presentation and report to take as much time as they have, and as such we’ve had to place some tasks on the backburner. We also still have not received our force sensors, and we plan to check in with receiving during the week to see if there have been any shipping issues. However we are still making good progress and we plan to use the slack time built into spring break to catch up on our various implementations.
Our deliverables for the week are a completed design report, a more customized model to do CV detection with, processed gyroscope (and force sensors if they arrive) data, and a working test of I2C between boards.