Completed our Intrim Demos this week, spending a lot of time to get the system working and presentable. Fixed some parts of the gantry which were coming loose after modification. Focusing on ML model and CV refinements to distinguish between pieces for games like chess. Also focusing on integrating the gantry with a table and tabletop to have a more standard board game playing experience.
Extra Questions:
- For our verification and testing phase, we want to focus on consistent accuracy over the course of a board game’s piece movements. For testing, we will have multiple users of different backgrounds play the three board games we support, and they will report whether each of the movements made by the gantry is accurate or not accurate in a survey. Along with this, they will also report the accuracy of the CV in detecting their moves correctly in the same way. We want to ensure the enjoyment and playability of FlexyBoard as described in our use-case requirements.
The subsystem I worked on was the microcontroller code for the stepper motors. We have found that slower gantry movement leads to more accurate movements, and we adjusted the speed to have an arbitrary balance of accuracy and speed. Based on the tests, I will be able to either increase the speed for faster movements if it is already accurate or decrease the speed to increase movement accuracy if it is too inaccurate.