by John Mack
For anyone who’s been paying attention to the robotics industry, there can be no doubt that robots are going to change the world. Over the past few years, we’ve seen enormous growth on both the hardware and software side; the robots of today make those of yesterday look rudimentary by comparison. One of the most fascinating ways in which we’ve evolved our machines is by teaching them how to learn.
The technology – known as deep learning – is arguably still in its infancy. Although powerful, its effectiveness is severely diminished when used on its own. Though we can teach our robots how to do things, deep learning doesn’t help them make logical inferences or draw on abstract knowledge. In order to carry out either of those tasks; in order to actually ‘think,’ we need to integrate a somewhat overwhelming array of algorithms and mathematical techniques.
It’s worth mentioning that to some degree, we’ve successfully accomplished this in the past. I’m sure most of you are aware of Watson, the supercomputer that defeated the previously undisputed human champion of Jeopardy. It was a moment of triumph for the scientists who worked on the device; finally, we had a machine that was truly capable of thinking and reasoning…right?
In truth, even Watson isn’t as smart as one might hope – the scope of what it can do is actually fairly narrow. It can’t, for example, participate in conversations about how it functions, nor is it equipped with facial recognition or the ability to pick up on emotional cues. Although I’m certain we might eventually see IBM integrate this functionality (if they see the need to), it’s still a troubling reminder of the current limitations of artificial intelligence.
Of course, all this talk about supercomputers is somewhat irrelevant to the world of robotics. Our robots have to be mobile – they have to be capable of working and moving alongside human beings. They can’t be a massive collection of servers, or a huge box of hardware. A housekeeping robot can’t be the size of the house it’s supposed to clean; a search-and-rescue drone can’t be stationary. As a result, we’re quite limited in what we can do with our robots.
That’s where cloud computing comes in. Hook up a robot to a cloud network, and suddenly it doesn’t matter so much what sort of hardware it’s equipped with in the physical realm. It has access to a potentially limitless sea of information, and could even tap into the processing power of a supercomputer while carrying out its daily tasks.
Let’s say, for example, you’ve got a robot that’s designed to cook meals for your family. While wandering around the kitchen, it encounters a utensil it’s never seen before – a pair of tongs. While an ordinary robot might either stop dead in its tracks or ignore the tongs altogether, a cloud-enabled robot could draw on the collective knowledge of thousands of others like it, identifying both the device and its intended purpose.
We’d basically have a collective database which every connected robot would contribute to whenever it learned something new.
The potential applications of the cloud don’t stop there, either. Tasks we humans take for granted – facial recognition, grasping objects, and simple movement – require an immense amount of processing power to carry out. Hooked up to a cloud network, a robot could gain access to that power without needing a metric ton of hardware. The end result is smaller, lighter robots.
Robots are going to change the world. This is an irrefutable fact. They won’t do it alone, either. Whatever the robots of the future look like, there can be no doubt that they’ll integrate the cloud in some way, either for collective knowledge, increased processing power, or both. After all, robots might be disrupting our world, but cloud computing is disrupting theirs.
John Mack is a technical writer for Datarealm, one of the oldest web hosting companies. You can follow Datarealm on Twitter, @datarealm, Like them on Facebook, and check out more of their web hosting articles on their blog, http://www.datarealm.com/blog.
Photo courtesy of Shutterstock.