This paper evaluates a system, which uses phone-based accelerometer, to identify a physical activity (walking, jogging, climbing stairs, sitting, and standing) a user is performing. Data is collected from 29 users, and this training data is used to make a predictive model for activity recognition.

In this presentation, we learned the working of accelerometer, which measures proper acceleration, associated with the weight of a test mass, and how we can use it for different applications.

Question answer session

In the discussion, the main topic was the applications of the accelerometer in the mobile device.

a) What are other applications of accelerometer in the phone?

Using activity recognition, we can gain useful knowledge about the habits of millions of users. Other applications include: customization of mobile behavior based on user activity, generating a profile to determine the amount of exercise, and patient recovery profile.

b) What will be the value of y if phone is placed vertically (portrait)?

The physical activity can be determined through accelerometer in three directions (x-axis, y-axis, and z-axis). The sign of y will be positive, because y-axis is controlled in that direction. X and Z are controlled in other two directions.

c) Can we do this activity recognition system in real time?

Yes, it is possible. But, this paper doesn’t cover in real time. In the future, it can be done.