It’s a given that we will have ‘intelligence’ in our cars within the next decade. Quite frankly, there’s no way it is avoidable, given the appetite of consumers for
all things to be connected to the Internet and too each other. In the case of cars, it actually makes sense for them to be talking to each other. But there’s one question: what will the unintended consequences be?
Earlier this week the National Highway Traffic Safety Administration (NHTSA) revealed plans to implement vehicle to vehicle (V2V) communication technology that allows one car to communicate with another and transmit information about location, speed of travel and direction of travel. Basically, 10 times a second a V2V car tell other V2V enabled cars its exact location, where it’s headed and how fast it’s getting there. The theory is that this would enable your car to warn you when someone is going to run the red light in front of you or is merging onto the highway in an unsafe manner near you. Presumably this would also integrate into smart car technologies, enabling them to better fend for themselves in high traffic conditions, since they’d no longer have to solely rely on their own sensors in the decision making process.
I have a host of security concerns about the idea of V2V cars, since most of the manufacturers who are creating the Internet of Things have shown that security is their last concern, if they even think about it at all. I can imagine the V2V system being used to track individuals every movement in a way that makes Orwell’s 1984 look Utopian. The privacy implications of having a car that’s constantly beaconing its location are pretty severe and in all likeliness the ability to track individual cars will be mandated by law. I can also imagine someone breaking into the communications systems to cause chaos, either by targeting an individual vehicle with false information or by disrupting a segment of the network that V2V relies on. At least there is someone else who’s thinking about the security concerns of interconnected vehicles, mainly I am the Cavalry and their Five Star Automotive Cyber Safety Program.
But what I find interesting in relationship to V2V is work that’s being done in swarm intelligence, as it relates to the idea of cars. Researchers at the Harvard School of Engineering and Applied Sciences have developed a swarm of tiny robots that can self-organize into a number of shapes without needing a central controller to manage them. The tiny little robots, Kilobots have very little intelligence (meaning computing power) individually and they don’t know much about their position as compared to the whole of the swarm, yet they manage to communicate with their peers in order to create organized shapes when they receive a command from the researchers. They know where they are in relationship to other robot near them and they use this information as to figure out what their role should be forming the shape requested, rather than having some sort of central program with an overview of the whole telling them what to do.
The swarm research that’s being done at Harvard is directly relatable to the V2V technology that (NHTSA) is doing. Even if there is never a centralized tracking program implemented with V2V (which I posit there will be, it makes tracking easier for the government) there will be swarm behavior from these smart cars. Swarm behavior already exists on our roads, it’s just that instead of a computer program making decisions, it’s human beings with limited awareness of the world around them. We make the same sorts of decisions that V2V cars would be making constantly; we call it ‘driving’. Most humans don’t have an overall view of the roads and what’s going on, though a lot of work has gone on to develop apps to give us this awareness of traffic.
Part of what makes a swarm of cars interesting, and a little scary, is the concept of emergent properties, or the idea that the whole is greater than the sum of its parts. This is exactly what’s going on with the Kilobots, the emergent properties of their intelligence means that the whole is able to figure out how to form shapes without an individual Kilobot having to be told exactly where it’s place is in the grand scheme. It’s up to the individual to do it’s best to conform to the needs of the whole to create the shape. But while the emergent properties of the Kilobots was the end goal of the experiment, what happens when you design a swarm of cars without an emergent property in mind?
We’re in the beginning stages of understanding how a swarm does what it does. How does a flock of birds really fly and wheel in unison? How does a school of fish form and stick together? How does a swarm of bees operate? Maybe over the next 5-6 years we’ll have a better understanding of what makes these things work like they do, but will this understanding be applied to our vehicles? The implications of a system of cars that have some sort of emergent property concerning how they enter, exit and move through traffic could be pretty severe, unintentionally creating gridlock and other safety concerns. It could also work to alleviate the same gridlock in unforeseen ways, which makes the technology worth pursuing.
And then there’s the sci-fi concerns, ala Maximum Overdrive. Swarm behaviors plus smart cars could create a series of emergent properties that make our cars decide that the safest option is to not get on the road in the first place. Or that it’s better to be in the middle of the swarm and keep driving instead of getting off at the proper exit. Or a hundred other scenarios that science fiction authors have explored in depth multiple times. It’s not that this sort of ending is a certainty, it’s more that it’s a possibility that has to be explored and prevented, rather than dismissed as an impossibility.