KSAS

KSAS (Kenpo Set Assisting System) is an AI Android application to learn martial arts movements in online settings.

The pandemic of the last years have forced thousand of instructors and learners to practice psyhomotor activities in online settings. The use of regular videoconference software and e-learning techniques has allowed them to continue their training at home. However, some problems specific to learning and teaching psychomotor activities have been exposed:

• It is difficult to visualize 3D movements in 2D displays.
• There is no physical interaction with classmates and teachers.
• It is difficult to give and obtain feedback to correct/reinforce the movements.

With the purpose of atteanuate those problems we have developed KSAS, an AI assisted Android application that helps to learn the order the order in which a set of movements is executed. The set of movements that you can practice with KSAS is Blocking Set I, a defensive set of movements taught in American Kenpo Karate.

KSAS is built over five different phases:

• Motion Capture: 3D inertial information of the movements is captured using the inertial sensors of a mobile device.
• Motion Model: The captured information is modeled as a time series and smoothed by using the EWMA (Exponentially Weighted Moving Averages) algorithm.
• Motion Analysis: A LSTM neural network is used to analyze the movements and determine which movement has been executed.
• Motion Response: The results of the analysis are given to the user in the form of haptic and auditory feedback.
• Motion Reports: The number of errors commited by the learner is stored, and a chart is created to show the progress of the learner.

KSAS is an open source application that follows the philosophy Open Science. The source code of the application is available in our Github website.