Fitness is one of the most paramount things in life and one of the way to keep track of fitness is to count the number of steps you've walked in a day. StepDetection app essentially gives you a way of tracking the steps at the comfort of your palm using the smartphone accelerometer.

What it does

It detects the steps traversed by the user.

How I built it

The current design uses multiple thresholds and peak values of accelerometer to calculate steps accurately. The values of the accelerometer are filtered first to remove noise, unwanted acceleration due to gravity and give value of forward accelaration which is perpendicular to gravity vector.

Challenges I ran into

One of the most challenging aspect of designing the app was to maintain accuracy of the steps counted. Initially while working on the app I decided to use the usual process of calculating the distance by double integrating the acceleration and hence counting the distance traveled. But on double integration the errors during the calculations were far too large and the steps were not reflecting properly.

Accomplishments that I'm proud of

This is my first time participating in the short hackathon series. Understanding the difficulties in building the app and going through many research papers to accomplish my task definitely was an awesome learning experience.

What I learned

I tried out various algorithms and approaches to tackle the problem of dead reckoning or inertial navigation. I realised where such algorithms failed. Although GPS is pretty good for navigation, but due corona virus and being restricted to my house, I learned its' limitations. Among such algorithms I felt steps tracking gave better accuracy.

What's next for StepsTracker

I plan on using the step detection funtionality to futher create path tracing app for inertial navigation and some more functionalities like Augmented reality path tracing.

Built With

Share this project: