Industrial and Commercial Robots

The first real-world use of robots had to do with manufacturing industries. But these systems did take quite a while to get adoption.

The story begins with George Devol, an inventor who did not finish high school. But this was not a problem. Devol had a knack for engineering and creativity, as he would go on to create some of the core systems for microwave ovens, barcodes, and automatic doors (during his life, he would obtain over 40 patents).

It was during the early 1950s that he also received a patent on a programmable robot called “Unimate.” He struggled to get interest in his idea as every investor turned him down.

However, in 1957, his life would change forever when he met Joseph Engelberger at a cocktail party. Think of it like when Steve Jobs met Steve Wozniak to create the Apple computer.

Engelberger was an engineer but also a savvy businessman. He even had a love for reading science fiction, such as Isaac Asimov’s stories. Because of this, Engelberger wanted the Unimate to benefit society.

Yet there was still resistance—as many people thought the idea was unrealistic and, well, science fiction—and it took a year to get funding. But once Engelberger did, he wasted little time in building the robot and was able to sell it to General Motors (GM) in 1961. Unimate was bulky (weighing 2,700 pounds) and had one 7-foot arm, but it was still quite useful and also meant that people would not have to do inherently dangerous activities. Some of its core functions included welding, spraying, and gripping—all done accurately and on a 24/7 basis.

Engelberger looked for creative ways to evangelize his robot. To this end, he appeared on Johnny Carson’s The Tonight Show in 1966, in which Unimate putted a golf ball perfectly and even poured beer. Johnny quipped that the machine could “replace someone’s job.”⁸

But industrial robots did have their nagging issues. Interestingly enough, GM learned this the hard way during the 1980s. At the time, CEO Roger Smith promoted the vision of a “lights out” factory—that is, where robots could build cars in the dark!

He went on to shell out a whopping $90 billion on the program and even created a joint venture, with Fujitsu-Fanuc, called GMF Robotics. The organization would become the world’s largest manufacturer of robots.

But unfortunately, the venture turned out to be a disaster. Besides aggravating unions, the robots often failed to live up to expectations. Just some of the fiascos included robots that welded doors shut or painted themselves—not the cars!

However, the situation of GMF is nothing really new—and it’s not necessarily about misguided senior managers. Take a look at Tesla, which is one of the world’s most innovative companies. But CEO Elon Musk still suffered major issues with robots on his factory floors. The problems got so bad that Tesla’s existence was jeopardized.

In an interview on CBS This Morning in April 2018, Musk said he used too many robots when manufacturing the Model 3 and this actually slowed down the process.⁹ He noted that he should have had more people involved.

All this points to what Hans Moravec once wrote: “It is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility.”¹⁰ This is often called the Moravec paradox.

Regardless of all this, industrial robots have become a massive industry, expanding across diverse segments like consumer goods, biotechnology/healthcare, and plastics. As of 2018, there were 35,880 industrial and commercial robots shipped in North America, according to data from the Robotic Industries Association (RIA).¹¹ For example, the auto industry accounted for about 53%, but this has been declining.

Jeff Burnstein, president of the Association for Advancing Automation, had this to say:

And as we’ve heard from our members and at shows such as Automate, these sales and shipments aren’t just to large, multinational companies anymore. Small and medium-sized companies are using robots to solve real-world challenges, which is helping them be more competitive on a global scale.¹²

At the same time, the costs of manufacturing industrial robots continue to drop. Based on research from ARK, there will be a 65% reduction by 2025—with devices averaging less than $11,000 each.¹³ The analysis is based on Wright’s Law, which states that for every cumulative doubling in the number of units produced, there is a consistent decline in costs in percentage terms.

OK then, what about AI and robots? Where is that status of the technology? Even with the breakthroughs with deep learning, there has generally been slow progress with using AI with robots. Part of this is due to the fact that much of the research has been focused on software-based models, such as with image recognition. But another reason is that physical robots require sophisticated technologies to understand the environment—which is often noisy and distracting—in real-time. This involves enabling simultaneous localization and mapping (SLAM) in unknown environments while simultaneously tracking the robot’s location. To do this effectively, there may even need to be new technologies created, such as better neural network algorithms and quantum computers.

Despite all this, there is certainly progress being made, especially with the use of reinforcement learning techniques. Consider some of the following innovations:

• Osaro: The company develops systems that allow robots to learn quickly. Osaro describes this as “the ability to mimic behavior that requires learned sensor fusion as well as high level planning and object manipulation. It will also enable the ability to learn from one machine to another and improve beyond a human programmer’s insights.” ¹⁴ For example, one of its robots was able to learn, within only five seconds, how to lift and place a chicken (the system is expected to be used in poultry factories).¹⁵ But the technology could have many applications, such as for drones, autonomous vehicles, and IoT (Internet of Things).
• OpenAI: They have created the Dactyl, which is a robot hand that has human-like dexterity. This is based on sophisticated training of simulations, not real-world interactions. OpenAI calls this “domain randomization,” which presents the robot many scenarios—even those that have a very low probability of happening. With Dactyl, the simulations were able to involve about 100 years of problem solving.¹⁶ One of the surprising results was that the system learned human hand actions that were not preprogrammed—such as sliding of the finger. Dactyl also has been trained to deal with imperfect information, say when the sensors have delayed readings, or when there is a need to handle multiple objects.
• MIT: It can easily take thousands of sample data for a robot to understand its environment, such as to detect something as simple as a mug. But according to a research paper from professors at MIT, there may be a way to reduce this. They used a neural network that focused on only a few key features.¹⁷ The research is still in the early stages, but it could prove very impactful for robots.
• Google : Beginning in 2013, the company went on an M&A (mergers and acquisitions) binge for robotics companies. But the results were disappointing. Despite this, it has not given up on the business. Over the past few years, Google has focused on pursuing simpler robots that are driven by AI and the company has created a new division, called Robotics at Google. For example, one of the robots can look at a bin of items and identify the one that is requested—picking it up with a three-fingered hand—about 85% of the time. A typical person, on the other hand, was able to do this at about 80%.¹⁸

So does all this point to complete automation? Probably not—at least for the foreseeable future. Keep in mind that a major trend is the development of cobots. These are robots that work along with people. All in all, it is turning into a much more powerful approach, as there can be leveraging of the advantages of both machines and humans.

Note that one of the major leaders in this category is Amazon.com. Back in 2012, the company shelled out $775 million for Kiva, a top industrial robot manufacturer. Since then, Amazon.com has rolled out about 100,000 systems across more than 25 fulfillment centers (because of this, the company has seen 40% improvement in inventory capacity).¹⁹ This is how the company describes it:

Amazon Robotics automates fulfilment center operations using various methods of robotic technology including autonomous mobile robots, sophisticated control software, language perception, power management, computer vision, depth sensing, machine learning, object recognition, and semantic understanding of commands.²⁰

Within the warehouses, robots quickly move across the floor helping to locate and lift storage pods. But people are also critical as they are better able to identify and pick individual products.

Yet the setup is very complicated. For example, warehouse employees wear Robotic Tech Vests so as not to be run down by robots!²¹ This technology makes it possible for a robot to identify a person.

But there are other issues with cobots. For example, there is the real fear that employees will ultimately be replaced by the machines. What’s more, it’s natural for people to feel like a proverbial cog in the wheel, which could mean lower morale. Can people really bond with robots? Probably not, especially industrial robots, which really do not have human qualities.