Title:

User Physical Activity Identification

The U.S. military has always been at the forefront of technology. As Machine learning (ML) becomes more and more prevalent, its applications in defense are rapidly explored by companies like Galvion. The purpose of our project is to create a neural network model that can recognize a soldier’s current physical activity, which will be delivered in the form of a Java library. Our work will be added to Galvion’s larger Heads-up-displays(HUD) ecosystem, where it will be used to inform the system what information to display. We hope the HUD will provide real-time support for the service members in time of need, and help overall mission success. Initially, we collected data from sensors on an Android smartphone secured around the chest of multiple test subjects, as well as, an external sensor module on the Smart Helmet provided by Galvion. To maximize the precision of our model, we experimented with different variables and how to record these variables. We ultimately found the optimal conditions to build our model around, which included using the accelerometer values, and timestamps, in addition to other variables. By using our model, Galvion is able to get real-time feedback for their HUD with minimal computing power.

Name

Comment*

Comments are viewable only by submitter

Captcha*