Many industries, such as construction, manufacturing, and service sectors, face talent shortages. Employers are having difficulty hiring workers, and this is true for low-skill, lower-wage jobs (and even semi-skilled and skilled job categories). Talk to anyone who has tried to get an electrician on-site lately, and you will know what I mean.
Conceptually, there are three principal ways to address this: immigration policy, education and workforce development, and robotics and automation. I’ll stress that all of these are important, can work in concert, and offer different tradeoffs and types of solutions. Changes to immigration policy are outside my realm of expertise—though, on a personal level, I have a mother who immigrated to the US, and I have worked with many, many talented people who’ve come to the US from other countries—I do believe that favorable immigration policies can offer a solution to critical worker shortages across many industries and allow continued economic growth in the US and other developed nations. I’ll also cover education, workforce development, and labor marketplaces in a separate piece. So what about robots?
Robots have long been heralded as a force multiplier for many jobs, whether addressing dull, dirty, or dangerous jobs. Examples include a range from Roombas to landmine removal. Or it’s in going places that humans can’t go: from the bottom of the ocean to inside the human body to outer space. If we are going to mine asteroids and the moon, the future is less likely to be a techno-thriller with roughnecks in space and more likely to be robots doing the real work. Perhaps the NASA Psyche mission is just the beginning…
Bringing back precious metals from asteroids aside, there are also ample opportunities right here on planet Earth where robots are poised to make a bigger impact and help drive productivity growth in many areas of society.
Now, stop me if you’ve heard this before. In fact, this was the premise that first got me excited about the field of robotics 15 years ago. Robotics as a field is right where the personal computer industry was in the 1980s. Computers at the time were big, bulky, expensive, and used mainly by industry, businesses, and hobbyists.
The span of a few short years in the early 1980s brought us the personal computer: IBM PC, Sun Microsystems, and the Apple Macintosh all emerged. We went from tens of thousands of devices sold per year in 1980 to tens of millions by 1990. The computer went from requiring a high degree of expertise to use to enter the home and the classroom and many other parts of the economy.
Robotics are at a similar inflection point.
In a very short time, image recognition has gone from relying on filters and edge recognition to using neural networks that perform better than humans.
Sensing and perceiving the world has gotten cheaper. The cost of a camera that will record high-definition video has decreased by 100x. Other sensors, such as Lidar—which provides 3D information about a scene—have followed similar cost curves.
Reinforcement learning (RL), a subset of machine learning, has emerged as a particularly valuable paradigm for robotics. Robots can now better move outside the predictable factory floor and operate in increasingly complex and chaotic environments, a la the real world. With RL, robots are better able to adapt to and learn from new situations. Robots can now even be trained in and learn from simulated environments—not dissimilar to pilots being able to train on a flight simulator— to decrease their learning curve in the real world significantly.
All of these technological advances and cost declines of the past decade have laid the groundwork for the next era of robotics. As many of the components have become commoditized, systems engineering and thoughtful integration of different parts are paramount for engineers building these systems.
On the demand side, labor only continues to become more expensive and difficult to find in many industries.
Safety becomes crucial as robots become more and more capable and move from relatively fixed, static environments, like a factory floor, to noisier, more chaotic environments, like a warehouse bustling with people, a construction site, a hotel, or a restaurant.
It’s extremely challenging (dare I say impossible even) for machines to have the versatility and ingenuity of humans. Still, they are getting better and can operate in increasingly noisy and chaotic environments. This trend will only continue to improve, and we expect to see much more in the way of collaborative robotics (that is, machines working alongside people) paving the way for safer and more productive paradigms.
Business models have evolved as well, and a conscious approach is needed for startups to focus on going from selling hardware to solving a customer problem. And the bottom line remains: it always comes down to cost. Thoughtful approaches look to minimize total operating costs, including time and expense incurred in setting up the robot in the first place.
Companies bringing robots into the world need to remain mindful that the best technology doesn’t always win. At the end of the day, it’s about solving a mission-critical problem, and often in the least complicated way possible. It’s better to be clever than it is to over-engineer the solution.
