Robotics Manufacturing: The Rise of Japan

“To the Americans, a robot is a computer attached to a mechanism. To Japanese, a robot is a mechanism attached to a computer.”

The future of the world economy will be shaped either by the rapid deployment of robots or by disappointment that robotics fails to improve nearly as rapidly as boosters suggest. Either way, the trajectory matters. I’ve been digging into the history of robotics to help understand this. There were multiple robotics “races” in the early decades of the robotics industry. One, which I’ll examine later, was the race to install as many robots as possible. A second was the race to build the best robots. In the 1980s, Japan was seen as a leader in both.

Though robots were invented in the U.S., Japanese firms played a major role in the U.S. robotics market by the 1980s. Most of the robots the U.S. imported came from Japan.

Concern over import dependence led the U.S. government to commission a report in 1983 on the competitive position of the U.S. robotics industry.

The report had the following explanation for America’s reliance on imported robots:

Imports made substantial inroads into the U.S. robot market at an early stage in its development. Early generations of domestically-produced robots were typically complex in design, expensive, and required frequent maintenance. In contrast, robots being produced in Japan and Europe were mechanically simpler, less expensive, and therefore easier for users to justify as capital expenditure items. The small number of U.S. firms marketing robots found it cost effective to import basic robots, adding a variety of peripherals, software and engineering in order to tailor systems to meet the needs of domestic users. Much of the U.S. robotics industry is still tied to foreign sources of supply for its basic product. The majority of U.S. robotics firms operate under some type of agreement with a foreign partner—joint venture, product manufacturing, and/or technology transfer. Some firms import completely assembled robots; others bring in components for assembly, adding ancillary items such as vision systems, end-effectors, and software in-house.

The largest U.S. robot manufacturers in the mid-1980s were Cincinnati Milacron (a traditional machine tools manufacturer), Unimation (which had pioneered the first robots), and GMFanuc, a joint venture between GM and the Japanese robotics firm Fanuc. As of 1982, Unimation was owned by Westinghouse, which itself used many of the company’s robots. GMFanuc mattered because GM was the country’s largest user of robots, so the more robots it purchased from GMFanuc, the smaller the market for other, purely domestic players.

The rise in imports coincided with a general deflation of the robotics bubble of the 1960s, marked by excessive optimism about the trajectories of robot capability and deployment:

Several competing suppliers [competing with GMFanuc] have been forced to move into less profitable lower-volume market segments, and have seen a reduction in their overall sales volume and market share as a result. At present, the U.S. robotics industry is supported by a substantial level of technology transfer from foreign robot producers. Many U.S. firms have purchased exclusive or non-exclusive marketing rights from Japanese and European robot producers, or have obtained licenses to manufacture these robots in the United States. Others have entered in some type of joint venture or technology exchange arrangement with one or more foreign firms.

As a result, U.S. robot imports had surged.

Most of these imports, as noted earlier, originate in Japan. Japan established an early lead in the world robot market with some very basic hardware that has been upgraded over time into a relatively complex technology. Until recently, the emphasis in Japan has been to put in place, as quickly as possible, as many industrial robots as Japanese industry could be induced to absorb. This strategy caused the Japanese market for industrial robots to grow dramatically.

Japan didn’t only benefit from a large robotics installation base which allowed economies of scale and innovation. It also specialized in different types of robots that were simpler and more practical:

Some of the price difference between the average U.S.-produced robot and its Japanese competitor can be traced to the difference in complexity of the robot mechanism itself. A large proportion of Japanese robots currently sold in foreign markets were originally developed for in-house use by Japanese manufacturers. Intended for a comparatively narrow range of applications, these robots are generally less sophisticated—and less expensive—than U.S.-produced robots. Overall, the special purpose mechanisms and manipulators favored by Japanese robot producers are less expensive to produce than the general purpose mechanisms on which many U.S. robotics firms have focused their development and marketing efforts.

Japanese robots were less AGI and more factory automation. And that is what the market wanted.

What about the future? The US government report noted ongoing U.S. advantages in underlying technology (my emphasis):

The potential for substantial advances in robot capabilities is perhaps greatest in the areas of sensors and artificial intelligence. Visual and touch sensors, which represent the most sophisticated stages of sensing technology, provide a robot with the “intelligence” necessary to recognize an object, determine its orientation, and inspect it for defects. Vision-equipped robots represent the third generation of robot development. U.S. producers have the potential to develop a decisive technological lead in this area. They will be directly challenged by Japanese producers, who have begun to focus their research efforts on robot vision systems and on autonomous robots that can operate without human intervention.

The U.S. could still hope for technological leadership in robotics because 60% of the cost of robot development was software—a sphere where the U.S. was widely believed to hold a major lead.

Indeed, a fascinating 1988 book titled Into the Robot Kingdom: Japan, Mechatronics, and the Coming Robotopia reported that Japanese firms believed that U.S. robotics companies retained technological advantages, especially in software, despite smaller production volumes. For example, Kawasaki licensed its first robots in 1968 from Unimate, the U.S. robotics leader. Two decades later, in 1985, it was still licensing tech from U.S. robot firms.

Adept Technology, a small California firm founded by former Unimation employees… Employing the latest in computer vision and American software, the Adept robot shows how American firms can still have an advantage over Japan in state-of-the art technologies, and also how thoroughly intertwined the Japanese and American robotics industries have become.

The point about the US and Japanese robotics industries being deeply intertwined is worth reflecting in on when thinking about industrial “leadership” then and now.

The first experimental direct-drive robot arm was developed in 1981 at Carnegie Mellon University in the U.S, mainly by two Japanese scientists, Haruhiko Asada and Takeo Kanade. In 1984 Adept manufactured the first commercial direct drive robot for light assembly, and for its basic structure it used a SCARA design. The firm that developed and made the new generation motors central to the Adept robot, and was for a time its sole supplier, was purchased in 1986 by a Japanese company, NSK, which moved all manufacturing facilities to Japan. Adept Technology today builds the motors under license from Japan and is the only US firm making direct drive robots.”

Forty years ago, like today, Japan was considered good at hardware and the U.S. was considered good at AI. The 1982 collaboration between Japanese robot manufacturer Fanuc and GM was seen as “a chance to tap the software expertise that Japan lacks in artificial intelligence and machine vision.” Fanuc’s CEO stressed “the superiority of American applications engineering, particularly in software, and suggesting the need for an international division of labor, where the U.S. handles software development and Japan the hardware.” Inside the Robot Kingdom reported that Japanese researchers study mechanical engineering while Americans study AI. One Japanese researcher was quoted: “To the Americans, a robot is a computer attached to a mechanism. To Japanese, a robot is a mechanism attached to a computer.”