A Brief History of Robotics since 1950

views updated

A Brief History of Robotics since 1950

Overview

The term "robot" comes from a Czechoslovakian word for "work" used in the 1921 play by Karel Capek called R.U.R. ("Rossum's Universal Robots") to describe an army of manufactured industrial slaves. Since then, we have come to think of robots as the mechanical men or "androids" of modern science fiction. In reality, technical manuscripts from as early as 300-400 b.c. reveal that human beings have been trying to build automated machines or "automata" for centuries.

The development of modern robotics was precipitated by the advent of steam power and electricity during the Industrial Revolution. A growing market for consumer products drove engineers to devise ways of producing automatic machines to speed up production, do tasks that humans could not do, and to replace humans in dangerous situations. In 1893 Canadian professor George Moore produced "Steam Man," a prototype for a humanoid robot made of steel and powered by a 0.5 horse-power steam engine. Essentially a gas boiler housed in what looked like a mechanical suit of armor, it could walk independently at a rate of 9 miles per hour (14.5 kph) and pull light loads. In 1898 inventor Nikola Tesla (1856-1943) demonstrated a model for a remotely operated submersible boat at Madison Square Garden. Tesla also wrote that he believed it possible to someday build an intelligent, autonomous humanoid robot. Tesla's ideas were not taken seriously until well into the twentieth century. In fact, the robotics industry as we know it emerged only around the mid-twentieth century. Once research and development teams began to work in earnest, however, robots were integrated into manufacturing and gradually adapted to the military, aeronautics and space, medical, and entertainment industries.

Background

By the 1950s engineers were developing machines to handle difficult or dangerous repetitive tasks for both defense and consumer manufacturing—particularly the booming automotive industry. Because robots were meant to replicate the pattern of movement that a human would make while lifting, pulling, pressing, or pushing, designs were based upon the anatomical structure and movement of a human arm. These were modified versions of the first patents for robotic arms filed over a decade earlier. For example, patents for both the "Position Controlling Apparatus," filed in 1938 by Willard V. Pollard, and a spray-painting apparatus by Harold A. Roselund, filed in 1939, were modeled on human shoulder-arm-wrist configuration and dexterity. Roselund's design patent was granted to the DeVilbiss Company, which would later become a major supplier of robotic arms in the United States. These early prototypes were not mass produced. However, once electronic controllers came into use after the Second World War, similar but more efficient designs were developed, including the first computer-controlled revolute arms from Case Western Reserve and General Mills in 1950, and a complex, hydraulically powered robotic arm by the British inventor Cyril W. Kenward, who filed his patent in 1954 and published it in 1957.

"Planetbot," one of the first commercial service robots in production, was a hydraulically powered robotic arm first used by a division of General Motors in the production of radiators during the mid-1950s. Eventually, approximately eight Planetbots were sold. The company claimed its robot could easily perform 25 individual movements and could be reset to perform a different set of operations in only minutes. However, this early model proved unsuccessful because it was controlled by a cumbersome mechanical computer, and it behaved erratically when the hydraulic fluid was cool. By the 1980s the Planet Corporation had developed a more sophisticated and efficient hydraulic arm, which has been successfully used for forging operations.

"Unimate," designed by George Devol and patented by Devol and Joe Engelberger, was originally used to automate the production of television picture tubes. The movement of Unimate's 4000-pound (1,816 kg) arm was controlled by commands stored on a magnetic drum. In 1962 it was integrated into General Motors Corporation production to sequence and stack hot, die-cast metal components. After the bugs were worked out of its design, Unimate became a popular feature in assembly lines. Of the approximately 8500 machines originally sold, more than half of them were used in automotive manufacturing plants. Today, there are approximately eight models of Unimate available, boasting payloads of from 50 to 500 pounds (23 to 227 kg). They have been adapted to such applications as material handling, spot welding, die casting, and machine tool loading, with an advertised 98 percent reliability rate.

Some of the most significant early contributions to robotics were sponsored by agencies outside of consumer manufacturing. The Case Western Reserve arm developed by Norman F. Diedrich was supported by the Space Nuclear Propulsion Office. A similar design, the "Programmable Universal Manipulator for Assembly" (PUMA) developed in the 1960s by Victor Sheinman, a graduate student at Stanford University, was intended for microsurgery. PUMA was eventually licensed and improved by the Unimation Corporation in 1978.

Taking advantage of advances in other technologies, developers gradually integrated more sophisticated computer controls and precision components into their models. Higher degrees of freedom—how far a component could move away from "home" position—led to greater versatility. The addition of finger joints in some models created manual dexterity for grasping, holding, and positioning objects. In 1960 a General Electric research team headed by Ralph Mosher developed "Handyman" and "Man-Mate"—two remotely operated robotic arms. Handyman, a two-arm electro-hydraulic teleoperated robot, had two hinge-type jointed fingers that could grasp objects via a single command from the human operator wearing the Handyman apparatus. Other robotic "arms" were actually based on the human spine. The multiple joints in these "serpentine" arms allowed for flexibility required in product-inspection operations.

A different kind of robotic arm configuration was used in NASA's Viking mission to Mars during 1975-76. The Viking landers, designed by the Martin Marietta Corporation, had to be designed with the extreme environmental conditions of Mars in mind. Instead of the heavy, jointed industrial arm, the Viking arms were made of two light, ribbon-like extenders rolled onto a drum. The two halves unfurled and connected, creating a tube to scoop samples from the planet's surface. Although the arm control mechanism had some bugs in it, operators on Earth were able to guide the robot through a repair procedure, making researchers optimistic about aerospace telerobotics—human control of robots from a remote location.

The desire to have machines that could follow commands or operate by themselves through many complex operations required special programming and controls for the machines. Some researchers came to believe that the best way to control what machines did for people was to find some artificial means to simulate the way that humans thought, remembered, and responded to their environments. Thus the study of artificial intelligence (AI) grew up alongside robotics engineering. The term "Cybernetics"—the study of the relationship between human and machine intelligence—was coined by scientist Norbert Wiener (1894-1964) in the 1940s, while he and a colleague, Dr. Arturo Rosenbluth, were working on ways to improve the automatic controllers in military aircraft. Since that time researchers have been experimenting with computer simulations of human thought. Although researchers disagreed about how the human brain works, the AI projects that arose from their debate have become a significant factor in the development of robots.

Another trend in robotics was to create mobile robots that could operate independently of humans. In order to do this, the robot must be able to avoid stationary and moving objects in its path. "Shakey," a primitive version of a mobile service robot based on this idea, was built at the Stanford Research Center in 1968. It had a vision sensor (a motorized camera and range-finder) positioned above a central processing unit (computer) on a cart. Shakey was propelled by two motorized wheels and two obstacle-sensing bumpers. It could apply logic-based methods of problem solving that allowed it to recognize the shape of objects, push them, and negotiate a ramp. Like Shakey, service robots of the 1980s were essentially "brains in a box" that ran along pre-mapped pathways laid out on the floor and could recognize and maneuver around obstacles. More recent models, like the RoboKent(r) SweeperVacs and Recycling ScrubberVacs from the Servus Robots Company, learn the area to be cleaned by themselves. They operate independently, without reference targets, and have builtin obstacle detection and avoidance protocols. Robots have cleaned more than offices: they have been used for larger, environmental hazard jobs, such as the cleanup at the Chernobyl nuclear power plant in the Soviet Union, following a major radioactive explosion there in 1986.

Shakey's ability to learn from interaction of its sensors with the environment also became the basis for the small, insect-like robots developed by researchers such as Rodney Brooks of the Massachusetts Institute of Technology (MIT) beginning in the 1980s. Brooks believed that the interaction of machine sensors with their environment creates a learning situation for robots similar to that of a human infant. According to this theory, it is not necessary to build complex computers with thousands of stored facts to control the robot. Instead, simple motor and sensor elements are combined to create robots that learn from experience. Brooks built robots modeled on multi-legged insects such as "Genghis"—designed to negotiate the kind of rough terrain that would be encountered on other planets. NASA planned to utilize "Hermes," a smaller version of Genghis, to explore the surface of Mars.

Many researchers argue that a humanoid robot is best for adapting to the human environment. When Rodney Brooks and Andrea Stein co-founded the COG humanoid robot project at MIT, their goal of producing an android that could behave like a human being and interact with human beings was considered controversial. Soon, other similar projects were in development. Vanderbilt University School of Engineering has been working on a humanoid robot called "ISAC." ISAC, like COG, is learning to interact with human people in a natural way. The projected use for ISAC is to perform as an in-home care-giver. Other university-based robotics research in the United States, including projects at the University of Southern California, the University of California-Berkeley, and Georgia Tech are destined for use in medical, private and military security capacities, or in environmental hazard situations. U.S. and European robotics research is tackling issues such as mobility, robot-human interaction, vision systems, speech imitation and recognition, and cognition.

These topics are nothing new to Japanese robot designers. Japan, which has almost twice as many robots as the United States, is developing android type robots for wide-ranging uses—to work outside of space stations and spacecraft, to interact with its growing geriatric population in hospitals and at home, and to act as civil servants in urban centers.

By 1986 the Honda Motor Company had completed "P-1," meant to "coexist and cooperate with human beings." Honda expected that by the time their humanoid robot was perfected, such robots would be used in everyday life to serve humans, not just in special operations. In the fall of 1997 Honda completed "P-3," which looks like a suited astronaut. It has a backpack with a 136-volt battery, wireless receiver, and processing unit. Commands are transmitted by the wireless ethernet modem. According to Honda, P-3's vision sensing system is able to identify stairs and other objects in a room, walk up stairs, and restabilize itself when pushed off balance.

At Waseda University, Japan, researchers have been working with another android project, "Hadaly." Like MIT's COG, Hadaly still looks much like a Leggo project with cameras for eyes, but researchers in other labs are working hard at simulating a human brain to someday operate inside of androids like Hadaly. Meanwhile, Fumio Hara and a team of researchers at the Science University in Tokyo have been working on an industrial "face robot" that can identify dozens of human expressions and make facial gestures itself. Recently, the face robot has been fitted with a simulated skin, hair, and eyes. Its designers believe that worker interaction with a humanoid face that provides emotional as well as verbal response will help reduce industrial accidents.

Impact

As with technology in general, corporate and industrial developers and independent inventors alike have enthusiastically adapted robotics to the entertainment and public relations industries. Corporations like Sony are beginning to market robotic pets that look and behave like cats or dogs, for people with allergies, or those who don't have the time to take care of a real pet. Since the 1980s novelty robots have appeared at trade shows, conference openings, and in safety programs at grammar schools.

Robotic technology has been integrated into in every facet of our lives, from manufacturing to military strategies, medicine, and other public and private service industries including environmental cleanup, space and underwater exploration, and entertainment. The integration of artificial intelligence (AI) and robotics during the last quarter of the twentieth century has resulted in predictions that androids—autonomous humanoid robots—will be a part of our everyday lives before the end of the twenty-first century.

LISA NOCKS