Archive for May 9th, 2010
F1 Data Junkie – Lap Elevation Data
Fans of cycling will be more than familiar with the idea of a profile map that details the elevation above sea level along a particular race stage.
(See also More Thoughts on Data Driven Storytelling for several more ideas on representing geospatial time series augmented by other sensor data.)
Although F1 circuit maps are provided by the FIA for each Formula One Grand Prix, they don’t give elevation data. Nor did a quick web search turn up any elevation maps.
In several previous posts on the topic of visualising F1 telemetry data I’ve plotted various map views, so I wondered whether I could generate elevation maps too… and it appears I can, using the Google elevation API:
The x-axis is distance round the lap, so if data from multiple laps is captured, we can start to get a more complete set of altitude data round the circuit.
Simply pass the API one or more sets of latitude/longitude co-ordinates, and I can get back elevation data. So for example, the above data (captured from the Vodafone Mclaren Live website) shows a lap by Jenson Button of the Malaysia circuit earlier this year.
So here’s what I’m thinking: how about a set of telemetric circuit guides derived from the telemetry data?
Here’s what we’d be competing with – the FIA circuit maps:
To do this, I think it would make sense to use data captured across the whole of a race, putting it into meter wide sLap bins (or maybe moving average bins 3 meters wide?) and then recording for each bin:
- the average latitude and longitude, to try to generate some idea of racing line;
- the most common (mode) gear setting
- the most common “g-force” values to show directional forces on the car; (or maybe we could derive an angle of travel based on current and next, or current and previous locations?);
- the average (mean) speed (maybe with outliers removed?)
- some average of brake and throttle values?
As to how to display the map – the use of Google elevation data requires that a Google map is also displayed, so using something like this map/scatterplot mashup technique might be appropriate?
Having good (average) resolution data for lat/long around the circuit as a whole also means we should be able to generate a reasonable KML tour to view a lap animation in Google Earth (e.g. reusing this Google Earth path/tour simulator (Silverstone data, F1 car kmz model)).
The only thing is – I can’t get myself motivated to hacking the code required to do this today:-(
PS hmmm… seems like racing line data is avaliable on the Racecar Enginerring website (e.g. Formula 1 2010: Round 5 Barcelona tech data). There’s also this interesting article on GPS data – it’s just a shame that the resolution we get on a per lap basis from the Mclaren site is at too poor a resolution (one sample per second) to be able to do anything really interesting with it…
