Objective
We wanted to provide a low to no cost solution to learning to type not on the screen. Since the device is in landscape mode the screen real estate used up by the keyboard provided limited options and we were strategizing to have an evolution of paper products that we could then upsell to a physical product.
We wanted to provide a low to no cost solution to learning to type not on the screen. Since the device is in landscape mode the screen real estate used up by the keyboard provided limited options and we were strategizing to have an evolution of paper products that we could then upsell to a physical product.
Evolution of the Keyboard
During the process of the paper keyboard I was also developing the hardware version of the keyboard, so at first we were looking at keeping the keyboard as closley related to that. I was also making many adjustments based on computer vision performance. I had also experimented with slide gestures on the paper and moving keys to reduce blocking vision. In the end we were able to use the BYJU'S logo and side markers to replace the ArUco marker.
The Final Printable Keyboard
Usability Testing
I ran many usability tests and gathered as much data as I could with where the children were struggling and where we needed to make AI improvments and UI improvements.
I ran many usability tests and gathered as much data as I could with where the children were struggling and where we needed to make AI improvments and UI improvements.
AI Data Collection
We were targeting two locations India and North America. The India team was doing training and testing but we needed a bigger data set to cover skin color and various lighting conditions. So I was conducting usability testing as well as collecting data to help train the AI. We developed a routine to follow to help progress the AI.
We were targeting two locations India and North America. The India team was doing training and testing but we needed a bigger data set to cover skin color and various lighting conditions. So I was conducting usability testing as well as collecting data to help train the AI. We developed a routine to follow to help progress the AI.
Learn to Type Application
We had already defined a style that was being used in the BYJU'S Lab eco system, where this application was going to live. However in the app we wanted to give it more of a "learnining path feel." Almost like Super Mario World. I help guide the direction of the feeling of the app and collected comeptive data and defined what we wanted to address in the beta to the final product.
Pop the Balloon
While I was investigating other apps, I built a HTML 5 prototype of what I had believed was the most appealing way to identify the key and cause a reaction with the child, and did a small samples test with lots of positive results.
While I was investigating other apps, I built a HTML 5 prototype of what I had believed was the most appealing way to identify the key and cause a reaction with the child, and did a small samples test with lots of positive results.