A big part of why GPS is so popular with customers is the ability to report on speeding vehicles. We offer real-time speed alerts with thresholds you can define, along with boundaries (e.g. > 50 mph in this construction area, > 75 mph everywhere else).
Here is an example of one which a customer wanted us to verify for them yesterday [they claimed the vehicle was done moving at 2:30 PM and couldn’t have been going that fast anyway in rush hour traffic].
I pulled that vehicle’s history up, and saw it had been moving, and probably just accelerated quickly up, then back down, from an average of 63 mph to 82, and then back down to 41 mph when the point was actually sent (we send instantaneous, average speed, and max speed every two minutes).
Here is a snapshot of that information, along with the 2.38 miles I show the vehicle moved in that 2 minutes since the prior point:
Doing the math, 2.38 miles in 120 seconds equals 71.4 mph average. Taking into account the fact the vehicle had slowed to 41 mph when it reported, and the fact that the average speed was 63mph, intuitively it makes sense that the vehicle had to have sped up at one point in order to cover that distance in only 2 minutes.
But is that enough to PROVE the vehicle was speeding? Since we get our speeds right off the engine’s computer, they’re highly accurate. We do get an occasional bad speed reading from some vehicles’ computers [literally 1 in 200,000 by my estimates], however, so we still try to sanity check speed alerts. Rather than break out my statistics books, however, let’s look at this vehicle’s recent history:
We can do that by only dialing in the last half hour or so of activity before the supposed speeding infraction of 82mph at 5:10 PM
This shows our map with ONLY the points from 4:40 to 5:14 visible, but the day’s path is still shown in light-to-dark, thin-to-thick blue (to make it intuitive by itself):
It helps to know that LIGHT GREEN dots (vs. regular green) indicate speeding > 66 mph (adjustable), and this vehicle’s recent history shows a lot of them. Clicking on one for more detail shows a max speed of 81 mph between 4:56 PM and 4:58, with an instantaneous speed of 75. Clicking on a few more show the speeding activity is consistent for that period of time, which would refute the “it’s 5:00 PM rush hour, I couldn’t possibly have been going that fast” argument:
Depending on circumstances, a max speed of 82 mph isn’t really horrible. Many freeways have 75mph speed limits, and accelerating to avoid an accident or to safely merge or change lanes is not something to discipline a driver over. But with a GPS system, 99.99% bullet-proof speed/location reporting, AND the tools to quickly react (alerts) and validate (maps/reports) this data, you’ll be able to know exactly what you want about your drivers’ speeding habits without too much difficulty. And trends definitely start to emerge for a number of our customers’ drivers.
The next article I think I will analyze this using some statistics from my all-but-forgotten 9 statistics classes I took in college. A friend’s new book How to Measure Anything got me thinking about how we can quantify whether or not single-point samples such as this are able to be used to prove speeding behavior. Doug Hubbard, the author, and I knew each other in the Army and have talked frequently about these types of measurements. I’ll see if I can enlist his expert help to better provide our customers with statistical mechanisms for identifying problem behavior in their drivers.