Looking for input from others to work out the logic to actually make the associations for the TPMS readings.
The busier your street is, the harder it is going to be to do it accurately. In the most simplistic case with very low traffic, we can just use the timestamps and determine that if we got TPMS data within 30 seconds or so of a plate recognition, the sensors likely belong to that vehicle.
As the traffic increases, things get much more complicated. Especially considering the fact that we might get a TPMS reading from a car as far as about 50 feet away. It is possible to manually trigger a TPMS sensor to transmit, but I'm not exactly sure how it works. If anyone thinks they could figure it out and wants to read the spec for the sensors, that would be super helpful.
Devices like below can "wake up" a sensor by transmitting something very close to the sensor. That means that it's possible to do it from farther away with more power. This might be worth exploring if it could narrow down the zone we expect vehicles to be in when we receive the data.
View attachment 216929
Even still, it's a tricky problem. The solution will likely need to store multiple possibilities with varying levels of confidence and then refine the database through process of elimination over time. Some sort of elaborate guessing algorithm or way to try to make accurate guesses would make it a lot better.
With the most primitive possible reduction logic, it's still pretty helpful to have a list of sensor ID's that were nearby at the same time as the ALPR recognition. If you get TPMS off a car with no plate, you will at least have a set of plates (probably never more than like 20) that it could be.
Could be an interesting use case for a super basic custom AI model to go an estimate vehicle association probabilities at some point.
