Pocket Sensei
By Lilo & Hazel
Purpose
Many Dojos are currently closed due to the COVID pandemic, leaving Martial Arts students without in-person instruction. [1][2] Practicing at home, students can check their own static stances much easier than they can check whether they are keeping the correct form during movement. Our app will help students check that they are maintaining their form throughout movement.
How it Works
Our algorithm inputs acceleration data on movement in a stance and output recommendations on how the student could improve their movement. Right now we have created the algorithms for front stance and made the code to detect "good" or "bad" movement for all three dimensions. It then provides feedback/suggestions based on what it detects.
Student starts app
Student takes at least 10 steps in stance with phone in back pocket
App analyses data
App gives student feedback
Analysis
Future Steps
In the future we would like to expand our algorithm to correct other stances and to add identification algorithm to identify what stance is present using eigenvectors. If the algorithm could provide real time feedback it would also help students as they could fix their stances as they go. Last, but certainly not least, because of the current situation we were only able to calibrate this data on one student, so collecting data from other students and adapting the algorithm them is needed. This would also help us make our algorithm more perceptive, as with more data it could not only distinguish between "good" and "bad" but could also say how good or bad those stances were.