I Spy, iRobot
A bright, colourful, cyberpunk, futuristic robotic world.
I want to kick off the new year by talking about robots. Now, yes, I know that robots have proved endlessly disappointing for years and years and years. Up with flying cars, they’re probably the most disappointing technology. For me, personally, more so! A few years back, I was in Tokyo and missed seeing ASIMO by like 30 minutes. Story of my life.
It turns out that building robots is hard. It’s not that scientists and engineers haven’t tried; it’s that the combination of hardware and machine intelligence has proved vastly trickier than even the greatest technology minds suspected.
But there have been successes in robotics: we all love the latest Boston Dynamics video showcasing Spot; driverless cars such as Tesla are pretty much AI robots even if we are some way from Level 5 automation; ports and distribution centres are stocked with autonomous robots. However, if we want to really look for robotics progress, we must look to Asia, even the International Federation of Robotics (IFR) says so: “73% of all newly deployed robots were installed in Asia, 15% in Europe and 10% in the Americas.” That’s not because Asia is playing catch-up. No, no. In every year since 2012, Asia has had more robot deployments than the rest of the world. It’s that Asia is the future.
Maybe the time has come for robots to finally live up to the hype. There are a few reasons why the time for robots is now: we have widespread 5G connectivity and cloud infrastructure to deploy robotics everywhere. We can pair robots with digital twins, allowing robots to work in less controlled, more uncertain and dynamic environments, like offices and cities.
But more important is how we programme robots. Old fashioned robot programming was deterministic. A robot had to be programmed to move from A to B or to put one foot in front of another. This worked for precise and narrow deployments, which meant that robots were as inflexible as a block of ice on a freezing day. Again, the good people from the IFR know about this: “Programming and integration account for 50-70% of the cost of a robot application… AI could ultimately halve the time required to programme a robot, as well as significantly cut the time needed to re-task it.”
The safety controls (“stop if a human is within 5 feet”) are still hard coded, but computer vision algorithms that surpass human capability and the availability of datasets to train AI models on have opened up a whole new way of programming robots. We’re in the early stages of artificial general intelligence, and robots with AGI are a dream.
I crave the day when we have human and robot harmony, the sort of thing we saw in iRobot (only the first half of the film before the robots attacked and tried to take over). But this pipe dream may no longer be a pipe dream. There’s a metric called robot density, which is the number of industrial robots installed per 10,000 manufacturing workers. And it’s increasing in the right direction (albeit a bit slowly). Simply put, the higher the robot density, the more intensive the use of robots. There are now 3.9 million industrial robots; South Korea has a robot density of 1,000, Japan 399, Germany 397 and China 322 (2021 figures). The US is at 274 if you were wondering. 46% of the world's industrial robots are produced in Japan, so big up Japan!
So, have robots been disappointing? Maybe not? Maybe we’ve been looking in the wrong place. In manufacturing, robots are everywhere: electronics, freight ports, semiconductor factories and, of course, the daddy of industrial automation, the automotive industry. Robots are hard in environments where the action space, all the things that a robot needs to be able to do, is massive. How do you go about programming a robot to “run away from the lad in Stone Island gear”? You can with AI. So, consumer applications are hard, but what about service applications? We’re already starting to see it in restaurants and airports.
But I predict that it’s not going to be too far in the future that, just as we now build automotive factories to be robot first, it’ll be the same in offices; we’ll see robots in the military and law enforcement, and we’ll see robotised cities. Naver Labs, the Korean tech company, is making great strides to build service robots. Hongik Ingan! But it might be the cities which become the home of a bright, colourful, cyberpunk, futuristic robotic world.