Researchers are developing a new algorithm capable of tracking a moving target which could have huge implications for the future of transportation.
The future is a busy, busy place. The human population is increasing every day, and advanced technology is seeping into every crack of the modern world.
We’re finding new solutions to old problems all the time, like getting a cab after a long night out. These days it’s as easy as pulling out our smartphone and calling for a service like Uber, or Lyft, and waiting in a designated place for them to come pick us up.
It’s a simple system, but it has its drawbacks.
The driver is only human, after all, and they may have a hard time doing so despite the modern conveniences of GPS systems. Plus, you have to wait in a designated spot, which may or may not be uncomfortable.
I, for one, don’t like that last part. Even if the driver can’t find me, I have to stay put lest the whole system breaks down and I can’t get home.
Let’s not overlook the fact that I’m usually nice enough to actually stay in the same place. I can only imagine what it’s like to be a driver when your fare has run off on you.
Can you imagine what it’ll be like when autonomous cars get on the road to pick people up, and the people aren’t where the AI expects them to be?
Sounds like a logistical nightmare.
This begs the question: How can you keep up with a target that doesn’t stay in one place?
Acquiring a Moving Target
To solve that problem, we turn to AI software, the best data cruncher of our new techno-future.
Researchers at IBM Research – Ireland are hard at work developing a new moving target algorithm that can find a target wherever it is, which is pretty convenient. The target no longer has to worry about being in a predestined place at a planned time.
The solution was presented in a research paper entitled “A Scalable Approach to Chasing Multiple Moving Targets with Mulitple Agents” to the International Joint Conference on Artificial Intelligence in Melbourne, Australia.
Hopefully the presentation will be posted here soon.
The approach is solidly built on the back of a lot of previous research by IBM. That means that it is fine-tuned enough to boast a maximum delay time for the service it can perform.
Take the cab example from earlier. Let’s assume that the cab is an autonomous vehicle, and while you know when you want to be picked up, you don’t know where. With IBM’s algorithm, it wouldn’t matter where because there will be an AI that can dispatch a cab with directions to get to wherever you are.
In fact, ‘dispatch’ is a great word here. Think of the algorithm as an AI dispatcher, or an operator. And with that maximum delay time, you’re guaranteed to see that cab within, say, 30 minutes or less.
This kind of tech would work wonders for an autonomous cab service, but that’s far from all it can do.
The potential benefit of this technology is actually pretty staggering, and it could provide a lot of solutions for AI programs designed to interact with the real world.
Hitting a Moving Target
This moving target algorithm could provide a lot of solutions, but what can it be used for other than dispatching a fleet of Uber-AI?What kinds of things would you use an AI tracker for? #findmykeysClick To Tweet
Think big; This kind of AI could help organize a method of smart city transportation and help avoid traffic jams and delays. It could also serve as the link between software that can diagnose medical patients and the human that treats them.
The AI could also help rescue efforts, giving relief to those in a crisis. If you read our recent article on Hurricane Harvey then you saw the level of rescue efforts that were going on in Houston during that time. They were phenomenal (I should know, I had a front-row seat), but if an AI like this could have made things better I think that everyone would welcome the change.
Securing a Moving Target
All of this is, admittedly, a bigger step than I otherwise would have thought for AI. Who knew that giving an AI the ability to track something or someone in the real world had so many uses?
I have a ton of questions at this point. Chief among them is this:
What could software like this mean for our capability to track individuals?
All of the benevolent uses for this kind of technology sound great, but there’s always a dark side. Ideally, we should be aware of any kind of privacy concerns before software like this goes commercial.
For example, if this gives AI the ability to track or locate people, could it help law enforcement find criminals? Check out this article where this is already happening.
Could it help drones follow people, or check in on them? All of that would be great.
Could it also give people the ability to plant a RFID device somewhere so that a drone can track someone without their knowledge or consent? Could it be used by governments or organizations to spy on people no matter where they go? All of that would be not so great.
For now, the answers to these questions and more are in the research phase, but that shouldn’t stop us from asking them.