Team Status Update for Mar 1

We have no significant changes to our schedule and the risks of our project mostly have remained the same. This week, we worked on the design presentation and report, as well as our individual tasks. For the design submissions, we created a new system architecture diagram that specifies what we are building vs what is off-the-shelf (see above). We also hammered down the requirements for power, story cohesion, and user satisfaction.

On the ML side, the synonym generation is proving less accurate than expected (see Ashika’s post for more details), so to mitigate poor performance later on, we may integrate multiple machine learning synonym models (perhaps using SpaCy) or we can use thesaurus.com, but this will add to the latency. For now we will continue to use just NLTK, but come back to this issue once the FitBERT algorithm is built so we can see how this lower accuracy will affect the whole story generation with actual user examples.

For the robot, development has been delayed slightly because we got the SD cards for the Raspberry Pis this week. However, to combat this delay, Abha worked on the CAD for the robot arm and frame, which didn’t require the Raspberry Pis, and will work on the displays this upcoming week. This flip in the schedule made it so that there are no significant delays to the schedule.