In his recent post in Wired magazine, Matt Simon declares 2017 the year that “…robots really, truly arrived.” While I appreciate Mr. Simon’s optimistic proclamation about the state of robotics, as a practitioner in this market, I must respectfully disagree.
Mr. Simon begins by citing as evidence of this robots delivering food in cities, self-driving cars “swarming the streets”, and bipedal robots walking out of the lab into the real world, even doing backflips. Granted there have been impressive demonstrations of robots in the past year, but let’s take a closer look at these examples. Delivery robots have indeed navigated city sidewalks, but they have been carefully accompanied by a human that helps them in tricky scenarios. So while it shows tremendous progress, it is not yet really a totally autonomous process. Savioke’s Relay robots are perhaps a better example and have been delivering items in hotels, doing so already for the past couple of years. However, its operational environment has substantially fewer variables than traveling at street level and still requires work to integrate it into a typical installation. It is a considerable leap to assume that sidewalk bots are going to get there as quickly.
As for self-driving cars, also we certainly have seen significant accomplishments here in the advancement of technology. I have personally been a passenger in a Tesla that was driving by itself on a highway. But we should not consider the advancement of driver-assisted driving to mean that we have arrived where we can rely purely on cars to drive us unassisted through all traffic scenarios. City streets offer a much more challenging scenario. Here again, we should not conclude that certain functional capabilities mean that the task is done, any more than one would assume that the first steps of a toddler mean they have mastered self-locomotion.
There are good reasons to hope for self-driving cars as we as humans have proven to be very fallible in this common skill. But I suspect we will require our robot cars not simply to match our ability, but to be exceedingly better before we will be willing to rely on them to operate totally without our supervision. Tesla’s autonomous driving mode still requires the driver to be ready at any moment to take over. While that may sound like we must be very close to self-driving cars, consider that John Deere has had autonomous harvesters and tractors for over a decade, but they still require a human to actively monitor and supervise them. Even if the technology was sufficient today, issues such as liability and many ethical concerns may still push the concept of totally self-driving cars out for at least another decade.
It may also not be the best way to think about robots, as this lends itself to the question of whether robots may replace jobs (or people). Technology has typically been at its best when conceived of as a way to augment human ability. It may be inevitable in doing so, that certain forms of human labor may become obsolete. How many companies employ huge secretarial pools any more. And we are reminded almost daily about how calling people can be automated. But such an impact is not limited to robots. Every technology since our beginning had an impact on what people do and how we relate to each other, with some functions going away and new ones being created.
As Mr. Simon’s suggested evidence that it was the year of robots because they were walking on two legs, that’s wasn’t new in 2017. Honda has been demonstrating Asimo for the past decade, including walking up and down stairs and Boston Dynamics has been dazzling us with a variety of forms as well. Here again, while I’ll admit that doing a backflip is an impressive new wrinkle, we need put this more into context. First, we are talking about a robot whose price tag likely exceeds $1m. We also don’t know how many failed attempts it took to get the video or how consistently Boston Dynamics can demonstrate this “valuable” skill. The DARPA Robotics challenge, using some of the same robots a few years ago, was a more practical test. But let’s say that there is value in robot doing a backflip and that Boston Dynamics’ robots can do this reliably 99.999% of the time. Robot arms, with the careful programming of their physics, perform impressive tasks hundreds, perhaps thousands of times a day and have been doing for years. If you placed an order with Amazon this holiday season, there’s a good chance that some of their robots helped get your order processed, a technology was created at Cornell that was demonstrated in 2005. I was frankly more impressed with Rethink Robotics’ design that enables their robot arms to be operated safely around people and that technology is now also several years old.
There is no question that robotics technologies are advancing. And indeed the cost reductions for sensors and actuators as Mr. Simon writes have played an important part in this. Certainly, a sensor that once cost Sarcos $250,000 and now only $8,000 helps make robots more affordable. But $8,000 for a single sensor still puts it somewhat beyond the range of being likely to be incorporated into typical consumer technology.
That’s not to say that others will not. We at Hoaloha Robotics would not be able to develop our robot without the increasing availability, affordability, and capabilities of new sensors. But that trend will still need to continue. And there may be limits on some required technologies. Let’s take a motor, a component that most robots will require multiples of. While more applications that use them help brings costs down, conventional motors are ultimately made up of things like copper and magnetic materials. Unlike software and electronic components, the costs here that cannot easily be reduced even in large quantities.
Finally, Mr. Simon cites appearances of consumer robots like Kuri. However, so far such robots are far from being household appliances. Certainly, we’ve seen a lot of attempts over the past year, but typically those companies underestimated the time and overpromised on expectations. The result is an amusing collection of devices, but about as useful as a Furby or Aibo. Even SoftBank seems to have changed their focus with Pepper from being initially promoted as an “emotional companion” to more of a commercial kiosk. That leaves Roomba, and its now many clones, as still the only truly successful consumer robot in the market today, and even they are still not as effective as a manual vacuum cleaner.
As a veteran of the PC era, most of today’s consumer robots seem more like where things were in the late 70’s when it seemed amazing to have a computer that sat on your desktop. Those machines merely foreshadowed what was to come but did not really, truly announce the advent. It wasn’t until 1982, after IBM entered the market, that Time declared the PC, the “machine of the year”.
That said, what we have seen over the past year or two is the growing importance of changing how we relate to technology. With PCs, we were first willing to subject ourselves to interaction that required us to use our fingers over a device that was designed to avoid the mechanism of a manual typewriter. Then came the mouse, that required us to relate movements of our hands to move an image in two dimensions on a screen. With mobile devices, we came even closer to how we naturally interact with the physical world. Microsoft attempted to expand that with body movement, but with limited success. But what robots are helping to reveal is that we prefer the same kind of interaction with have with each other that includes multiple dimensions, including how we communicate with our voices and faces and must consider not just what is said, but how. However, this, too, did not really arrive in 2017. We are only seeing the early signs.
Mr. Simon wasn’t wrong in citing numerous examples of how robotics technology is becoming increasingly accessible. I would not be doing what I am doing if I didn’t share that opinion. But I’d say robots haven’t really, truly arrived any more than we’ve seen in the past decade. We certainly moved in the right direction in some important ways, but I would not say we are got there last year.