An Imitation of Life: Intelligence and Robots

Renato M.E. Sabbatini, PhD


What is a Robot?

The word robot comes from the Czech word robota, meaning "forced labor". It is one of these words, created by science fiction, that catch on. It appeared for the first time in Karel Çapek's 1920 play R.U.R, where robotic workers -- "mechanical men" -- rebel against their masters and assume control of the world after slaugthering them.

The fact is that, from a technological point of view, we are very far from having a robot with enough skills, intelligence and autonomy to replace human beings at the majority of tasks and chores.   The robots that exist today (and in large numbers) are industrial models. However, we can hardly refer to them as intelligent. The sole thing they are capable of doing is to reproduce sequences of highly complex movements such as holding, displacing, releasing, pinpointing, touching, pulling, and so on, imitating a human arm with all its "degrees of freedom", that is, how many joints does it have, in which directions each one could move.

At an automobile assembly line, for instance, or at an electronic printed circuits plant, where repetitive and precise processes take place, the robots perform well ... until something gets out of the programmed sequence or positions (this is the case if a board falls down or a component happens to be slightly smaller!). Then comes disaster, since artificial robots, with rare exceptions, are not intelligent, that is, they do not possess the senses of vision, touch or hearing, do not how to "feel" the object or solve an unexpected problem, nor have adaptative algorithms which automatically adjust themselves to the completely new situations.

To understand the enormity of this capability, it is enough to imagine, for example, how extremely difficult it would be to build an industrial robot to perform such a simple task as to break an egg and scramble it over a pan, up to the point it can be served. For this, the robot would need a sense of vision (to see the egg); tactile sense (in order not to smash it by excessive pressure when the egg is picked up); the sense of propioception (sensors that indicate the position of arms and fingers); as well as the delicate control required to break the egg exactly in the middle of its shell and over a frying pan (which needs to be at the right temperature) and so forth.

Several researches made at the advanced robotic centers in the USA, Europe and Japan are managing to create models of future robots with a minimum degree of "intelligence". For instance, a Japanese laboratory create a robot with vision that is capable of reading a musical score and interpret it over the piano. A German university developed an autonomous robotic vehicle that "navigates" through a room full of obstacles, thanks to a visual system composed of three eyes (the German scientists believe this "unbiological" number of eyes renders more easily the acquisition of 3-D scenes of the environment). Another company is already marketing the first robotic hands that have the sense of touch and automatically regulate the pressure exerted over the objects they handle.

Another robotic sensation was the Sojourner rover (mobile robot on wheels) that landed in Mars in 1998 and was remotelly controlled from Earth. As soon as it escaped from the delivery vehicle on the harsh surface of Mars, little Sojourner began to explore the surrounding terrain and examine its rocks. This was surely something that called the attention of newspapers, magazines and television, and robots, which were somehow forgotten, returned suddenly to center stage. Also in 1998 a high-tech Japanese company launched a robot that remarkably resembled a walking astronaut. The robot can move up and down stairs, run on flat terrain, avoid obstacles, and so on, in an autonomous manner (that is, it is not necessary that a human being remotely controls the robot, as is the case of Mars Pathfinder). Another Japanese company generated a big frisson when it begun selling $2,000 robot-dogs, capable of several apparently affectionate and intelligent behaviors and sold them out in a few hours.

Despite all these efforts, an intelligent robot does not yet exists. Of course, there are many project for flexible and autonomous robots. The specialists use the word "adaptive" to describe a device that is capable of modifying its actions according to environmental changes. For instance, if a car's door that needs to be soldered is slightly out of position in relation to the ideal parameter, current industrial robots are incapable of correcting the situation by themselves without interrupting other ongoing tasks or causing a disaster in the production line.   Easier said than done. It means, technically, that the robot capable of this achievement needs to have a system of specialized sensor and that they should send complex information to the robot's brain (the internal computer) that controls the robots own movements, in order to as to enable some kind of "sensorial and mechanical coordination",. Such sensors, as it happens in human beings, can use several types of information: audio, video (mono or binocular vision), tactile sensors, and so on. In some robot projects, they are fitted with ultrasound sensors (a kind of sonar that allows the detection of obstacles around the robot) as well as laser sensors (to recognize 3-D objects). ALV or Autonomous Land Vehicle, which is the correct name of NASA's Martian robot, is equipped with an array of such sensors.

To write a computer program capable of combining all the possible "sensations" with the movements that can be performed by a robot can be extremely dificult and even so it couldn't contemplate all possibilites. It is enough to say that the Martian robot "crashed" several times, when one wheel got stuck up in the air, when it stumbled over a stone and could not get rid of it, and so on. The last problem was generated by a small flint caught between the Rover suspension and the wheels. If it were a living being, it probably would have paws, hands or a snout to get rid of the annoying rock. In the Rover case, it would stay completely blocked until someone does not find a trick to go around the glitch.

Intelligent Robots

As the genial W. Grey Walter predicted (and experimented with), the future of truly intelligent robots lay in the conjunction of Artificial Intelligence and Robotics, two disciplines which are already veruy much interwoven. A technology called "artificial neural networks" tries to imitate the organization and functioning of the brain by means of artificial elements that behave like networks of neurons (the basic cell of the nervous system), but using silicon microchips instead. Neural networks present the great advantage of learning complex tasks such as 3-D recognition of obstacles, controling movement of joint articulation in robotic arms, and so on.

Our nervous system perform well complex tasks such as correctly frying an egg because:

 The first robots that display such advanced technology have shown amazing flexilibility and adaptation capabilities. It is thought that an autonomous robot will have several hundreds of such little "brains", controlled by a major computer, and that  they will act in an integrated manner with sensors and motors, exactly as it happens with a real nervous system. Reflexes, intelligent decision-making and automatic systems will make the robots of the future very similar to what Isaac Asimov and other anticipated so vividly. What is difficult to guess is how long it's going to take.

Are Robots "Human"?

Now we can discuss again why people wished to adopted W. Grey Walter's turtles and even express tenderness towards them.

Sherry Turkle, a psychoanalist and researcher from the Massachussetts Institute of Technology (Cambridge, USA), who, in 1984, wrote a "cult" book called "The Second Self: Computers and the Human Spirit", provides the answer. The book is based upon a series of researches, that she, as a sociologist, undertook to study the feelings people expressed towards computers and robots. She surveyed several kind of informatics users and professionals and came to some revealing and incredible conclusions. One of the major findings it that human being do not possess the psychological conditions to differentiate between a natural mind and emotion (as observed in another human being) from their artificial simulation (as generated by a computer). This conclusion bears extremely important consequences to the evolution of Artificial Intelligence and Robotics, as they increasingly approach the mental capacity of a human being (it suffices to note the somewhat hysterical reaction of chess world champion Garry Kasparov, when he was defeated by an IBM computer...)

In one the many anedoctes she describes, a teacher of robotics at the MIT presented his students a small and primitive robot, that was sold as a "kit" by an American company at the time. The robot was capable of performing small tasks, such as walking, avoiding obstacles, grasping small objects with its only mechanical arm, and utter vocally small synthesised words (when the robot was turned on, for instance, it greeted: "Yes, Master !"). During several classes, students had the opportunity to see Mike in action (as they nicknamed the and small nice robot, very similar to R2D2 of "Star Wars" fame), and rapidly became emotionally attached to the robot. (the teacher make a point in dealing with the robot as if if itas a human being or a pet). All of a sudden, on purpose, the teacher pushed it down the table and the robot smashed on the floor, and stopped working. As a result, several students were profoundly shocked by the robot "death".

These facts came to my mind following the amazing fad caused by the marketing of Tamagotchis, little digital "beings" that behave like living beings, which were invented and produced by the Japanese company Bandai Digital Entertainment. They are born from small eggs (in fact, just an image in a Liquid Crystal Display (LCD) screen -- No wonder, Tomagotchi means "adorable tiny egg", in Japanese) -- , they emit signals when are hungry and need to be feed by their owners (by pressing a button), they show signs of pleasure when they "see" the owner, they sleep, wake up and so on. One button is used to "punish" the Tamagotchi once in a while. It can also evolves into one of twelve different creatures. The toy has already sold millions of units in Japan and other countries of South East Asia, Europe and the USA. They have provoked a rash of many interesting phenomena that demonstrate the emotional link that appear between man and machine, when it simulates rational living beings. According to some people who became owners of these cute creatures, they are funny to watch and to take care in the beginning, but gradually they become more and more "demanding" and irritating.

Just to give some examples of this emotional attachment, some Tamagotchis owners stopped working or studying just to "look after" their "babies". Tamagotchi's baby care centers were open, where for a small amount of money they "took care" of the "baby" when its ownwer had to travel or to work. Even Tamagotchis cemeteries have appeared (since they "die" after a certain period of time, by lack of food or by abandonment ..) as well as Internet home pages where the owner can declare his/her affection towards them or leave an epitaph. Some kids even underwent emotional shock when their little Tamagotchi "died".

Inevitably, the incredible human imagination went into creating thousands of varieties of Tamagotichis (in one such variety, two Tamagotchis can be coupled together to fight bloody battles that end when one of them dies). This lead us to imagine that in the future, we may witness the appearance of Tamagotchis "societies", with professional experts, housewives, children and adults, and so on that can interact one to another in many different ways, including involving real human beings.

 All this may seem astonishing to many people, but do not come as a surprise to the readers of Sherrry Turkle's book.

Copyrigt 1999 Renato M.E. Sabbatini
All Rights Reserved
First Published: July 25th 1999