At Ather, while examining data from the bunch of sensors on the vehicle I figured that this data could potentially capture a person’s ‘unique’ style of riding. I had accompanied the folks for several tests at a racetrack myself, and noticed that this made a pretty fun and engaging narrative. How did one person ride, how did he/she do compared to the others?
Bringing out such nuanced information from a multitude of variables - throttle, acceleration, deceleration, speed, gryos, etc. - was inherently complex, and a cool data visualization seemed like a nice way to tackle this. A few weekends and a lot of pretty pictures later, I built one up in Processing (A Java-based sketching library) that gave a neat, almost game-like replay of the ride as it was logged. This was unexpectedly complex as I had to write a pre-processing layer, along with some basic sensor fusion to make sure the visual looked nice and was informative. I also had to create custom classes for polygons that would tile up along track path (rounded edges!).
Visualization guide - width of chunks: speed, color: acceleration / deceleration intensity, tiny grey bars next to throttle: histogram of throttle positions (rotated vertically)
The result? A pretty seasoned, agressive rider:
A more ordinary, death-fearing guy:
(NOTE: The above images are screenshots from an interactive applet that would allow for scrubbing and pausing. I’m working on integrating a cool interactive bit using Processing.js soon ).
Naturally, this led to reasonably geeky comparisons within the team. However, was interesting to see the effect of style on range performance: efficient needn’t mean much slower. Also might have safety potential: making people more aware of their style relative to driving norms.
Marketing Event Spin-off
What started as an interesting exercise ended up getting incorporated into a couple of events that were run with early-adopters. Look out for yours truly, and the blobby snakes from above.