Hi there. Yes, I know, it’s been 2 years since I last posted something on my website. Mea Culpa, I installed Joomla, I’m not much for website design, been busy, that sort of thing.
So how’s the spouse and kids? Really?! Well congrats, I’m glad to hear that. Except for the part about the restraining order, that’s a shame.
…
But hey the real reason why I’m here is to talk about an idea that popped into my head while I was zoning out along I-85 when heading back from a conference earlier this week. Being a responsible adult and all I tried to keep it down to only the “prayer for judgment” speeds and not “Judge Dredd” speeds. This is easy enough with cruise-control until you suddenly realize there’s a car ahead of you or behind you and you’re not sure if it’s a cop or not. And you don’t have radar detectors because you’re a law abiding citizen, damn it!
It occurred to me that we have all the technology to build an open source surveillance system for tracking and identifying police cars on the road. Now doing something like this of course requires making sure everything is done legally, and that has to account for differing laws between states and counties. But let’s push that down further into the article and jump into a technical outline of how to do this.
(And yes, this is a way to turn universal surveillance by law enforcement back on them).
Back-End Systems
To start with the idea of a Collaborative Cop Identifier System (CCIS) requires servers and software to be able to pull in data from many contributors, process it and produce useful output. The CCIS would likely use a typical virtual hosting platform with servers for databases, webservers to handle automated and human requests, and app servers to run data processing tasks that are computationally intensive. Let’s assume relatively unlimited financial resources for this project, so everything can be hosted in a nice redundant datacenter (*cough*), we have paid coders to do dev work and sysadmins to keep everything running and monitor for outages.
CCIS would primarily be used by people driving cars, so it’d have to be optimized for use over WiFi/Cell networks. We’d need support for speech input and output, and a robust API so a lot of client platforms can be supported (cellphones, tablets, etc).
The core model would be individual users equipping their cars with a set of sensors, running a CCIS client application on an appropriate mobile device, pairing all this to an account on the CCIS backend systems and then just driving like they would normally. Data would be gathered by the client system and sent to CCIS servers for processing, and real-time feedback would be fed back in the form of voice commands, audio signals and/or visual cues.
Monitoring
Vehicles participating in the CCIS would need one or more sensors to assist in identification. The three best that come into my head are dash cameras, bluetooth antennas and the previously mentioned radar detectors. Dash cameras would provide a wide-angle camera view front and back and feed a simplified image stream into the CCIS back-end where various recognition patterns could run to try and identify police units. Bluetooth is a bit trickier as it would require a fairly large antenna to be able to read far signals, and research would be needed to determine what unique Bluetooth signatures a Cop car may provide (i.e. MAC address, identifier name). Radar detectors would provide a lot better information, but of course there’s a lot of false positives. But with an appropriate combination of these sensors, there should be enough raw data that a back-end system would be able to find multiple unique characteristics for a vehicle and correlate them.
In addition to those sensors there would of course be GPS support, as that would be necessary for correlation, pin-pointing locations, etc.
Collaboration
Now that we have a lot of raw data, how can we tell the cop cars from other cars? Actual patrol units would be pretty easy with direct image recognition as they have unique profiles. But undercover and unmarked units can be not just cars but trucks, vans and SUVs. So image recognition would require a lot more training, which is where the collaboration comes in.
By having a participant in the CCIS program vet whether a particular vehicle is, or is not, a cop car a better set of indicators can be built. For example, police fleets very significantly from region to region, so GPS data would likely play a pivotal role here (and with much of the system, for obvious reasons). Vetting could be done via voice responses to prompts, or a quick tap on a phone/tablet screen.
Part of the collaboration requirement would be that interaction need not distract a user from driving, so an emphasis on voice interaction would be necessary. For touchscreen interfaces, prompts would need to be very simplistic (such as flashing icon of a cop car with a super-imposed question mark) and response just as easy (tap the top part of the screen for yes, bottom part for no). Of course, after a trip a user should be able to login to the CCIS system via a more traditional website GUI and perform any ‘cleanup’ for incorrectly identified vehicles, delete trips you didn’t want tracked, etc.
Uses
The first and most obvious use of such a system is to avoid getting traffic tickets, which can be considered a morally questionable action, except for the widespread cultural acceptance we have for speeding and avoiding getting tickets. No one wants to condone the psychotic rampages of a person who books it 100+ MPH through a school zone and ends up having to be put down by a SWAT team. But those are rare cases having to do with violent, mentally questionable offenders, not Joe Average. Conversely, many speed-traps exist across our nation that unfairly trick people into thinking they are driving at a legal speed when they are not, to generate revenue for flyspeck towns.
Of course such a system could also be used by criminals to avoid pursuit, which means it is unlikely that any actual implementation of such a system will get very far before it faces legal challenges. Not to mention angering a lot of law enforcement officers.
However, when implemented correctly, CCIS could be valuable and useful for many people. Privacy concerns can be addressed by automatically blurring out facial images, license plates, etc. Google StreetView has shown that there’s a lot of value in this mass-surveillance activity and that while there are legal challenges, they are not insurmountable.
One last interesting aspect of this idea… If it was ever actually implemented and became widely adapted, we’d have a scenario where millions of people are constantly adjusting their driving speed based on how many police cars they encounter. In such a scenario, by simply increasing the presence of law enforcement on the road without actually increasing the number of ticketing events, a reduction in dangerous speeding can be achieved.
