An android programmer working at the intersection of cybernetics, psychology and behaviorism (behavioral science), and an engineer writing algorithms for industrial robotic complexes, whose main tools are advanced mathematics and mechatronics, are working in the most promising branch of the coming years - robotics. Robots, despite the comparative novelty of the term, have long been familiar to mankind. Here are just a few facts from the history of the development of smart mechanisms.
Iron Men Henri Dro
Even in the myths of ancient Greece, mechanical slaves were mentioned, created by Hephaestus to perform heavy and monotonous work. And the first inventor and developer of a humanoid robot was the legendary Leonardo da Vinci. The most detailed drawings of the Italian genius have been preserved to this day, describing a mechanical knight that can imitate human movements with his arms, legs, head.
The creation of the first automatic mechanisms with programmed control was launched at the end of the 15th century by European watchmakers. The most successful in this field Swiss specialists father and son Pierre-Jacques and Henri Droz. They created a whole series of humanoid robots ("writing boy", "draftsman", "musician"), which were controlled by clockworks. It was in honor of Henri Dro that in the future, all programmable humanoid machines began to be called "androids."
At the source of programming
The foundations for programming industrial robots were laid at the dawn of the 19th century in France. It was here that the first programs were developed for automatic textile machines (spinning and weaving). Napoleon’s rapidly growing army urgently needed uniforms and, consequently, fabrics. Inventor from Lyon, Joseph Jacquard, proposed a method for quickly reconfiguring a loom to produce various types of products. Often this procedure required a huge amount of time, tremendous effort and the attention of the whole team. The essence of the innovation came down to the use of cardboard cards with perforated holes. The needles, getting into the cut-out places, necessarily shifted the threads. Change of cards was quickly carried out by the machine operator: a new punch card - a new program - a new type of fabric or pattern. French development has become the prototype of modern automated systems, robots with the ability to program.
The idea proposed by Jacquard was enthusiastically used by many inventors in their automatic devices:
- The head of the statistical department S. N. Korsakov (Russia, 1832) is in the mechanism for comparing and analyzing ideas.
- Mathematician Charles Babbage (England, 1834) - in an analytical machine to solve a wide range of mathematical problems.
- Engineer Herman Hollerith (USA, 1890) - in a device for storing and processing statistical data (tabulator). Note: in 1911, the company. Hollerita was named IBM (International Business Machines).
Punch cards were the main media until the 60s of the last century.
What is a robot?
Intelligent machines owe their name to the Czech playwright Karel Čapek. In the play "RUR", which was released in 1920, the writer called the robot an artificial person created for heavy and dangerous production sites (robota (Czech) - penal servitude). And what distinguishes a robot from mechanisms and automatic devices? Unlike the latter, the robot not only performs certain actions, blindly following the laid down algorithm, but is also able to interact more closely with the environment and the person (operator), adapt its functions when changing external signals and conditions.
It is believed that the first working robot was designed and implemented in 1928 by an American engineer R. Wensley. The humanoid "iron intellectual" was named Herbert Televox. The pioneer laurels are also claimed by the biologist Makoto Nishimura (Japan, 1929) and the English soldier William Richards (1928). The anthropomorphic mechanisms created by the inventors had a similar functionality: they were able to move their limbs and head, perform voice and sound commands, and answer simple questions. The main purpose of the devices was a demonstration of scientific and technological achievements. Another round in the development of technology soon allowed to create the first industrial robots.
Generation after generation
Robotics development is a continuous, ongoing process. To date, three distinct generations of “smart” machines have formed. Each is characterized by specific indicators and applications.
The first generation of robots was created for a narrow type of activity. Machines are only capable of performing a specific programmed sequence of operations. Robot control devices, circuitry and programming practically exclude autonomous functioning and require the creation of a special technological space with the necessary additional equipment and information-measuring systems.
Second-generation machines are called sensible, or adaptive. Programming robots is carried out taking into account a large set of external and internal sensors. Based on the analysis of information from the sensors, the necessary control actions are generated.
And finally, the third generation is intelligent robots that are capable of:
- Summarize and analyze information
- Improve and self-learn, accumulate skills and knowledge,
- Recognize the images and changes in the situation, and in accordance with this build the work of their executive system.
The basis of artificial intelligence is algorithmic and software.
General classification
At any representative modern exhibition of robots, the variety of “smart” machines can amaze not only ordinary inhabitants, but also specialists. And what kind of robots are there? The most general and substantial classification was proposed by the Soviet scientist A.E. Kobrinsky.
According to the purpose and functions performed, robots are divided into industrial and research. The first, in accordance with the nature of the work performed, can be technological, material handling, universal or specialized. Research is intended to study areas and spheres that are dangerous or inaccessible to humans (outer space, earth's interior and volcanoes, deep-sea layers of the oceans).
By the type of control, biotechnological (copying, commanding, cyborgs, interactive and automatic) can be distinguished, by the principle - hard-programmable, adaptive and flexibly programmable. The rapid development of modern microprocessor technology provides developers with virtually unlimited possibilities in the design of intelligent machines. But an excellent circuit and constructive solution will serve only as an expensive shell without appropriate software and algorithmic support.
Robot Programming Basics
In order for silicon of the microprocessor to be able to take over the functions of the brain of the robot, it is necessary to "fill" the appropriate program into the crystal. Ordinary human language is not able to provide a clear formalization of tasks, the accuracy and reliability of their logical assessment. Therefore, the required information is presented in a certain form using the programming languages of robots.
In accordance with the tasks of management, there are four levels of such a specially created language:
- The lowest level is used to control actuators in the form of precise values of the linear or angular movement of individual parts of the intelligent system,
- The level of the manipulator allows for overall control of the entire system, positioning the working body of the robot in the coordinate space,
- The level of operations is used to formulate a work program by indicating the sequence of necessary actions to achieve a specific result.
- At the highest level - tasks - the program without detail indicates what needs to be done.
Robotics seek to reduce the programming of robots to communicating with them in higher-level languages. Ideally, the operator sets the task: “Assemble the vehicle’s internal combustion engine” and expects the robot to complete the task.
Language nuances
In modern robotics, programming of robots develops in two vectors: robot-oriented and problem-oriented programming.
The most common robot-oriented languages are AML and AL. The first was developed by IBM only for managing intelligent mechanisms of its own production. The second - the product of specialists at Stanford University (USA) - is actively developing and has a significant impact on the formation of new languages of this class. The professional will easily discern the characteristic features of Pascal and Algol in the language. All robot-oriented languages describe the algorithm as a sequence of actions of a “smart” mechanism. In this regard, the program often comes out very cumbersome and inconvenient in practical implementation.
When programming robots in problem-oriented languages, the program indicates not a sequence of actions, but goals or intermediate positions of an object. The most popular language in this segment is AUTOPASS (IBM), in which the state of the working environment is represented in the form of graphs (vertices - objects, arcs - communications).
Robot training
Any modern robot is a learning and adaptive system. All the necessary information, including knowledge and skills, is transmitted to her in the learning process. This is done both by directly entering the corresponding data into the processor memory (detailed programming - sampling), and using the robot's sensors (by the method of visual demonstration) - all movements and movements of the robot mechanisms are recorded in the memory and then reproduced in the duty cycle. Learning, the system realigns its parameters and structure, forms the information model of the outside world. This is the main difference between robots and automated lines, industrial machines with a rigid structure and other traditional automation tools. These teaching methods have significant drawbacks. For example, when sampling, reconfiguration requires a certain amount of time and labor of a qualified specialist.
A very promising program for programming robots presented by the developers of the Laboratory of Information Technologies at the Massachusetts Institute of Technology (CSAIL MIT) at the international conference of industrial automation and robotics ICRA-2017 (Singapore). The C-LEARN platform they created has the virtues of both methods. It provides the robot with a library of elementary movements with specified restrictions (for example, the grip force for the manipulator in accordance with the shape and rigidity of the part). At the same time, the operator demonstrates the key movements to the robot in a three-dimensional interface. The system, based on the task, generates a sequence of operations to complete the work cycle. C-LEARN allows you to rewrite an existing program for a robot of a different design. The operator does not need in-depth programming knowledge.
Robotics and Artificial Intelligence
Oxford University experts warn that in the next two decades, machine technology will replace more than half of today's jobs. Indeed, robots have long been working not only in dangerous and difficult areas. For example, the programming of trading robots has significantly crowded out human brokers on world exchanges. A few words about artificial intelligence.
According to the layman, this is an anthropomorphic robot that can replace a person in many areas of life. Partly it is, but to a greater extent, artificial intelligence is an independent branch of science and technology, using computer programs that simulates the thinking of "Homo sapiens", the work of its brain. At the present stage of AI development, it helps people more and entertains them. But, according to experts, further progress in the field of robotics and artificial intelligence can pose a whole series of moral, ethical and legal issues to humanity.
This year, at the robot exhibition in Geneva, Sofia, the most perfect android, announced that she was learning to be human. In October, Sofia was recognized for the first time in the history of artificial intelligence as a citizen of Saudi Arabia with full rights. The first swallow?
The main trends in robotics
In 2017, digital industry experts noted several outstanding solutions in the field of virtual reality technologies. Robotics did not stand aside. The direction that improves the management of a complex robomechanism through a virtual helmet (VR) looks very promising. Experts predict the demand for such technology in business and industry. Possible use cases:
- Management of unmanned equipment (warehouse loaders and manipulators, drones, trailers),
- Medical research and surgery,
- Development of hard-to-reach objects and areas (ocean floor, polar regions). In addition, programming robots allows them to carry out autonomous work.
Another popular trend is connected car. More recently, representatives of the giant Apple announced the launch of the development of their own "drone". More and more companies are expressing their interest in creating machines that can independently move along rough roads, preserving cargo and equipment.
The increasing complexity of programming algorithms for robots and machine learning makes increased demands on computing resources and, therefore, hardware. Apparently, the optimal solution in this case would be to connect the devices to the cloud infrastructure.
An important area is cognitive robotics. The rapid growth in the number of "smart" machines makes developers increasingly think about how to teach robots to interact seamlessly.