The Pursuit of Intelligent Machines: Part 4 – Here Come the Robots

Robotic devices or automatons have been around a long time. Intelligent human-like  robots, however, have not. We can find historical references to mechanical devices in Classical Greece, Ancient Egypt, early dynastic China, the early Abbasid Caliphate in Baghdad  and in the kingdoms of the Indian subcontinent. These early automatons were created to entertain and amuse.

When Mary Shelley wrote Frankenstein in the early 19th century, she described an artificial life form that was made up of human parts. It wasn’t however until the early 20th century that the term robot became part of our language. R.U.R. or Rossum’s Universal Robots was a story written by the Czech playwright Karel Capek. As in Mary Shelley’s story, humans invent the robots  with the robots becoming a threat to the inventors.

No individual popularized the term robot more than Isaac Asimov who in 1942 defined laws for robot behaviour.  The three laws countered both Shelley’s and Capek’s dark predictions of  destruction. Asimov’s laws of robotics stated that no robot could injure a human or allow a human to come to harm. No robot could disobey a human unless in following an order by a human, that or another human would be harmed. And finally, Asimov’s third law stated that a robot must protect itself from harm as long as in following this rule it didn’t injure a human or allow a human to come to harm.

In Stanley Kubrick’s 2001:A Space Odyssey, the movie portrays the ultimate fear in humanity’s pursuit of intelligent machines. Only the HAL 9000, the computer that oversees the Discovery spacecraft is aware of the mission’s objective and in keeping that information secure decides that the human crew accompanying it is a threat. The dramatic confrontation between man and machine is played out in the vacuum of space.

We have had robotic arms with us for more than half a century. These devices, designed to emulate and surpass the capability of the human  arm have been used for a wide range of applications, from industrial assembly to the International Space Station. A robotic arm features multiple joints with rotational capability and a grasping or mutli-purpose end tool at the tip. End tools can mimic a human hand or serve a specific purpose such as welding.

Today most automotive assembly lines use robotic arms extensively for the latter task. The first one of these, the Unimate robotic arm was installed at a General Motors facility in 1961. By 1973, Cincinnati-Milacron, a U.S. manufacturer, had developed commercial computer-controlled robotic arms that soon found themselves installed in hundreds of factories to provide precision welding and precise machining of parts. Seiko is just one of many Japanese manufacturers who have taken Cincinnati-Milacron’s invention and created smaller, single and multi-function robotic arms for the automotive, biotechnology, electronics, pharmaceutical and telecommunications industries. The CanadARM, CanadARM 2, and its successor, the Space Station Remote Manipulator System (SSRMS), successfully launched in 2002, are products of MD Robotics, a Canadian company. These devices have the ability to operate autonomously or be controlled manually from the Space Shuttle or the International Space Station where they are deployed.

But our ultimate goal is Data, the android of Star Trek, the creation of a machine that exhibits human traits, that appears human-like, and that develops a relationship with us.  The Massachusetts Institute of Technology Artificial Intelligence Laboratory is pursuing the creation of Data-like robots. COG is a creation of Professor Rodney Brooks, Director of the Lab. Along with a talented team of graduate students they have been building a humanoid robot. COG has been designed with many human-like traits from eye movement, head and neck orientation, facial recognition, and arm movement.  COG can even play the drums.

Another MIT robot is Kismet, a creation of a team led by Dr. Cynthia Breazeal. Kismet is an expressive robot that interacts with humans on so many levels. Although a disembodied head the robot expresses interest, sadness, calmness, surprise, happiness, anger and even disgust. If a person is out of range of Kismet’s immediate view but visible the robot engages in calling behaviour.  If someone gets too close, Kismet leans away.  Kismet both listens and talks to humans. Multiple computers are needed to create the combination of behaviour and attributes that are displayed by this very human-like robot.

The idea of building an intelligent robot in humanoid form is considered important because many scientists believe our body shape and structure govern to a large degree many aspects of our ability to express ourselves.  A humanoid robot, therefore, would interact with people like a human. MIT research shows that humans exposed to human-like robots interact with them as if they were human.

Japanese technology companies have been in forefront of creating humanoid robots.  Unlike the MIT robots, Japan’s research has focused on creating task-specific skills for a generation of automatons performing such functions as nursing, shopping, housekeeping and home care. Here are just a few examples;

HRP-4C is a female-shaped robot that walks, talks, and expresses emotions. A engineering project developed by the National Institute of Advanced Industrial Science and Technology, HRP-4C contains 30 motors to operate body movement and 8 to create its facial expressions. The robot has even been programmed to sing.

Saya is a work robot. Developed at the Tokyo University of Science, Saya has facial muscles that are pneumatically controlled allowing the robot to lift its eyebrows, crinkle its nose and smile.  Saya has been programmed to teach elementary shool children, act as a receptionist and do other general purpose activities.

Asimo is Honda’s humanoid robot. The first model appeared in 1986 and is probably the robot most of us know from video and film on Youtube. Asimo is marketed as a general purpose home ‘bot. It responds to voice, and human gestures. Looking very much like a space-suited tiny human, it can vacuum, shovel snow, play with children, serve food and refreshments and negotiate maneuvering through crowds. Asimo can recognize individual faces and greet people by name. The Asimo design has even developed technology to allow a human to control the robot through thoughts by the wearing of a helmet with sensors that detect the brain’s electrical impulses.

Twenty-One is a robot creation of Waseda University. It displays dexterity capable of picking up a straw, putting it into a glass and handing to a human.

AR, a Home Assistant Robot is a product of a joint venture among the University of Tokyo, Toyota, Fujitsu, Mitsubishi, Panasonic, Sega and Olympus. Standing 61-inches in height by 25.6-inches wide with a depth of 31.31-inches, and equipped with a wide-angle stereo camera, a telephoto stereo camera and ultra-sensitive sensors. AR operates on two drive wheels with balancing wheels. It can run on its battery for up to an hour and does such tasks as sweeping the floor, clearing and cleaning dishes, picking up clothes and putting them in the washer and moving chairs.

These anthropomorphic robots represent our first steps in creating the future Data. As we introduce new computing technologies such as biological-silicon hybrid chips, neurogrid  and quantum computing, and nano-circuits, our robotic devices will become more autonomous, more intelligent, and more capable. Machines that resemble us will become part of our daily existence here in the 21st century. Ray Kurzweil, the futurist who sees exponential growth in both the software and hardware capacity for us to create artificial intelligence, predicts that by 2029 robotic intelligence will be equal to human intelligence.

In our final blog installment on this topic I will discuss the blurring of machine and human intelligence as we and our machines develop a symbiotic relationship.