Data Collection At Breckenridge Village

On Friday November 6th, Project Silversteps traveled to Breckenridge Village, in order to show the progress that we have made since our last meeting/presentation with them. We began data collection after a quick review of why we were there and a explanation of the Samsung Simband, and a demonstration of our current progress on the hip connector accelerometer. We collected walking data from 6 volunteers. During the experiement, the volunteer would walk around the track for a given distance at his/her own pace. Meanwhile, Karly followed the volunteer counting their steps while Alex monitored and recorded the data from the Samsung Simband into SAMI. Each trial included the start and the stop time, the number of steps counted, the number of steps detected by the Simband and Garmin Fitness Band. We decided to take data from the Garmin so we could compare accuracy between the Simband and another popular pedometer that is currently on the market. Looking at the live feed through the SAMI website, the data collected was quite clean and will be very useful for our project. We were able to gather data on a wide variety of gaits: from slow to fast to with or without a walking assisted device. Project Silversteps will be returning to Breckenridge Village on December 4th to test and validate the hardware portion of our project and our final product.

From the data collected, it was clear there was a wide variety of motion detected. Those who used assiting devices to walk had a more defined steps and yielded a more accurate step count. While other styles of walking had a less defined step and gave a less accurate step count. Thus far, our team has 3 methods in our algorithm for counting steps. When running the SAMI data through the algorithm, it was clear that some methods gave a more accurate step count for certain walking styles. For this reason, we are considering analyzing the users walking style during the calibration phase and then based on this information assigning a specific counting method that will be more accurate for them.