Crowdsourced Traffic Cops

I have an idea… on how to punish stupid drivers. Rather then have a billion cops on the road trying to police every lane change and tailgate we can learn a little bit from the power of the internet. We will use crowdsourcing to identify bad drivers.

Here’s how it works:

First we add four cameras to every car —just like the cameras some cars have now for reversing— we will have one with a clear view in front of the car, one looking behind and one on each side. These four cameras will be on whenever the engine is running and they will record what they see to a storage device somewhere in the brains of the car, part of a in-vehicle control unit. The storage device will contain, let’s say, enough memory to record the most recent five minutes of video from each camera. If it’s not used (as below) in the five minutes then the video is just dumped.

Second we need a trigger for the driver. This would be some sort of easy to use button or switch on the dash or maybe on the steering wheel. When the driver activates the trigger the in-vehicle control unit will preserve the record of what is going on around the car by copying the current recording to a separate ‘holding memory’. But rather than taking just the previous five minutes the in-vehicle control unit will delay for two minutes 30 seconds before the recording is transferred. This means that the unit will capture what was happening on all sides of the vehicle for the 2 and a half minutes before the trigger was pressed and for the 2 and a half minutes after the trigger was pressed.

Once a recording is transferred to the holding memory a process would begin to send this information to a central processing system. How exactly this would happen would be a big hurdle but let’s say, for the sake of brevity, that there is some sort of ubiquitous network built into the road system such that the in-vehicle control unit can send the recording via a WiFi connection. We could imagine some additional information, such as gps trackers coordinates and a simple time-stamp being sent along with the recording, but we should refrain from sending any sort of data about the vehicle submitting the recording, it needs to be anonymous.

At the central processing system the video would be processed by a software system to identify all the license plates visible in the videos and ‘tag’ the videos in such a way that the system is able to retrieve the portion of any video containing a specific license plate.

The central processing system would calculate if the same license plate was showing up in a large number of videos in a period of time. If a license is showing up with some pre-set frequency it would be flagged for review by a human. For example if a particular license plate shows up in 5 videos in a 20 minute time span the various relevant video potions would be collected and sent for review by a human.

The human’s job in all this would be watch the videos and determine if the driver who’s license plate was flagged was doing something illegal or dangerous. Where they tailgating? Changing lanes erratically? If so they would be issued a ticket for this. And the videos could be used as evidence in any dispute.

The crowd sourcing is important, I don’t want a system that records everything blindly. I want the drivers of cars to use the trigger to flag behavior that they think is bad. If someone cuts them off or is tailgating them then they can just use the trigger and know that, if the person is habitually doing something wrong or behaving dangerously, they will be punished because others will also be sending in video of the bad behavior.

To avoid overly cautious people or vindictive drivers from abusing the system there would be some sort of daily limit on incident submissions. For example, you can only send in five incidents a day. And by requiring that there be multiple videos, triggered by multiple drivers of bad behavior the system relies on the wisdom of the crowd to know what is dangerous and undesirable behavior —one person cannot get you in trouble, everyone has to agree you drive like a moron— while still having a human on the back end, and expert or maybe a committee of them, making the final determination of if this is actually illegal and the driver is going to be punished. And of course all of this only works if every car is equipped with the system.

Sound a little big brother? Too many privacy concerns? Well, don’t worry I can’t imagine this actually happening in places like the US (well, maybe the UK… ironic huh?) This system is really for Singapore. It’s a good fit for Singapore; a small enough place that you could have the ubiquitous network —imagine the amount of coverage you’d need to cover the US? Singapore is like Manhattan. It also works because the Singapore government has the power to force things like installing the system in all cars. In fact they’ve already done something similar by mandating the fitting of an in-car stored value card reader/transmitter for the ERP system []. And most importantly the police don’t actually enforce driving rules here. The result of which is that everyone, despite having to know the rules to pass a strict driver licensing scheme, drives like they got their license from a box of cracker-jacks.

This is a much better alternative to my original idea, which was to arm everyone with a gun shooting little Nerf like flags that said “moron” on them and having the police stop and ticket anyone with 6 or more of these flags stuck to their (moving) car. Everyone gets a daily ration of 5 flags. That way you can shoot the morons who piss you off.

One reply on “Crowdsourced Traffic Cops”

Ha, as I was reading through this, I was thinking: “Man, he’s been living in Singapore for too long.” Then you admit it at the end. ;-)

For the record, these days most new cars here are fitted with a device that stores speed, traction information, etc on a regular basis. If the computer senses of impact or accident, it stores the information long-term — kind of like an automotive black box. It’s been used to prosecute people here in the last couple of years and there’s been a lot of brouhaha over whether it’s a violation of privacy, etc. They’re isolated cases, few and far between, but still…

Comments are closed.