Towards the end of this week, we had a conversation with Prof. Rowe about our capstone design, and he brought up some good points on how to better reach our objective. We ultimately decided to switch our design so that the hardware’s purpose is to support the detection algorithm; previously, we were needlessly focusing on the exercise of creating a single board computer without considering other design avenues.
Our new design will include modifying OctoPrint and running our algorithm on a RaspberryPi. Our hardware component will be to construct an RPI shield. However, we will be gathering data from 2 cameras and 2 lasers in order to build a better model of the current print and to provide better cross-checking. Our system will also, over its lifetime, build up a model of the distribution of actual print-errors and inherent errors built into the system (i.e. due to laser and camera calibrations) and assign weights, contributing to the total error that is used to determine whether a particular print is erroneous. Then, the user will be notified and asked if the print should be stopped.
Here’s an updated block diagram:
Because of the design change, we also needed to update our Gantt chart:
