Project Description:
Gait is the way that someone walks and is something that is unique to everyone, just like a fingerprint. We can measure someone's gait using an accelerometer. This allows us to map out a person's specific gait and using this data we can then predict the person's height.
Here is a link to the data that our class collected using an accelerometer:
https://docs.google.com/a/students.nusd.org/spreadsheets/d/150b1-yV3A_sxjVM_x5nBJOxQmjUJ6iwnHH4JuQfsgto/edit?usp=sharing
Here is a link to the data that our class collected using an accelerometer:
https://docs.google.com/a/students.nusd.org/spreadsheets/d/150b1-yV3A_sxjVM_x5nBJOxQmjUJ6iwnHH4JuQfsgto/edit?usp=sharing
Predictive Model:
For our predictive model we divided the step length by the average ratio to step length to height. To calculate the step length you take the distance walked and divide it by the number of steps. For our trial the distance was 25 feet in 10 steps so 25 divided by 10 equals 2.5. Then we divide that number by the ratio of step length to height which is .4. So 2.5 divided by .41 gets us about 6. This allows us to predict the height of most people and be within ±1 inch of actual height.
25ft/10steps = 2.5 foot steps
2.5 foot steps/.41 step to height ratio = 6
6 feet is height of person walking
25ft/10steps = 2.5 foot steps
2.5 foot steps/.41 step to height ratio = 6
6 feet is height of person walking
Reflection:
The major challenges in this project were to interpret all of the data collected. Each accelerometer was putting out hundreds of data values for a 25 ft walk, this map it very hard to synthesize the data into something easy to look at and understand. Another issue was that everyone in the class used slightly different methods of collecting data. The accelerometers were not uniformly mounted on the people walking making each data set slightly different and hard to interpret. What we were able to do well was communicate within our team and get everything that we needed to get done in time. I think that if we had better equipment and a more uniform experimental design it would vastly improve this experiment.