TelemVision

• 

  Driver Comparison at turn 7

• 

  Lap Comparison at turn 11

Inspiration

Telemetry data feeds can be complex and messy. TelemVision takes the hassle out of processing and interpreting this data by presenting it in a clean, clear and easy to understand way. TelemVision allows users to see what a driver was doing at every point in the race and see how they reacted to different features of the track. This tool can be used to develop pit strategies for drivers, advise how a driver can maximize grid position and defend against opponents by observing how competitors attack and defend through corners and straight aways.

What it does

TelemVision allows users to compare drivers performance from one lap to another or compare drivers performance head to head for any lap. Users can see what line a driver took through each turn, when they started braking going into a turn, when they started accelerating leaving a turn and how the car responded to these inputs throughout the race. The dashboard also allows users to zoom into specific sections of the race to get a detailed view of how the driver handled the car through specific turns or sectors. This tool is great for things such as comparing a driver's fastest lap to an average or slow lap to see how different line choices shaved time off the lap or how tire degradation impacted performance from one lap to another. This information can be used to inform pit strategy and influence a driver's line selection. Teams can also use this tool to compare their drivers with competitors to better understand competitors driving style and line choices and braking patterns.

How we built it

I wrote an R script to organize the data into a table with a each telemetry variable represented by a single column arranged by its timestamp. I then converted the lat long pairs into spatial objects and calculated the distance from the first point of each lap to the next and so on until each point had a new attribute which stored the distance traveled along the track. This step required me to select a local spatial projections to accurately measure the distance between each point. Once this attribute was assigned to all each data point I was able to visualize each telemetry variable vs the new distance traveled field and align the data spatially along the track. Additionally I used statistical interpolation methods to fill in missing values for some telemetry attributes which logged data at different intervals than the GPS unit. These imputed values provide a more complete picture of the race. Uncertainty of the imputed values is minimal due to the rate at which data was logged by other instruments and relatively small number of missing values. Once the data was cleaned I leveraged tableau to visualize the data and build a dashboard with a User Interface for the data.

Challenges we ran into

The main challenge I encountered was how to align the telemetry data from different vehicles. I did this by calculating the distance traveled from the start of each lap by converting the latitude and longitude point pairs from each vehicle and lap into lap traces (lines). I was then able to calculate the distance traveled from the first point recorded in a lap to the nth point recorded in a lap for each point in the dataset. With this new distance traveled along track attribute I was then able to visualize the telemetry data vs the distance traveled along the track and compare each driver's progress throughout the race.

Accomplishments that we're proud of

I am most proud of being able to clearly present the large amount of telemetry data in a clear and easy to understand interface which allows for post race analysis and insights.

What we learned

I learned about working with telemetry data feeds and manipulating spatial data as well as different methods of aligning telemetry data to allow for direct comparison of drivers performance between laps and compared to other drivers.

What's next for TelemVision

I would like to expand TelemVision to include telemetry data from all races as well as include telemetry data from qualifying. I would also like to build a new User Interface for TelemVision rather than relying on Tableau.

Built With

• r
• tableau