International Pioneer Infaiba Theory of biological evolution allows machine self-design, wisdom, this is a strange and difficult to predict. The United States Paul Lai Zel is an expert in an evolutionary hardware. As early as 3 years ago, he became a oscillator circuit with the transistor by the researchers of the University of Sustix, and several new circuits can meet the requirements of design functions. They have stable The output and a stable frequency. One of them is very strange, and it relies on a line that connects itself on the oscilloscope and the line. That is to say, it deceives its designers, it can be self-design, with partial features of life.
"My reaction to this is to be very sucking and amazed." Rai Zel said this, "Just as you see a child solves a common problem with a creative way, you will make a heart. "
Machines can evolve
Using evolutionary ways to improve the performance of the machine is not a new thing. Since the 1960s, artificially intelligent researchers use evolutionary theoretical dynamics to solve software issues related to financial investment, manufacturing and biochemistry.
However, it is still a new thing to apply evolutionary processes in hardware areas. Thanks to the new year-changing computer new technology, now designers can make their hardware products to evolve for a while so that they can see the results. Maybe some people question the evolutionary direction of such hardware products. However, for engineers, this test itself is enough to become a test of the 21st century technology.
Some people think that evolutionary projects are just a step in technology. But some people think that this will become an opportunity to break software and hardware innovation bottlenecks. Bird borrowed the concept of artificial intelligent researcher Brown University philosopher Danie Danite, "I will see this evolutionary radio to be a vivid thinking test, which can form a thinking model similar to us to solve the problem."
Useless circuits are like appendix
The Jim Engineering is a manufacturer specializing in antenna. It has begun to use evolutionary design as a military unit and the NASA manufacturing antenna. The company's scientific and technological person in charge of Derk Lin Deng will suddenly set off the evolutionary hardware design boom attribute to scientific and technological progress in the relevant fields. He said: "When I first started doing this, my simulation can only be in a Pentium I. Machine on the machine. Now you can simplify the problem easier. "
In evolutionary design, simplifying problems means decomposing problems into smaller, more single work. As the human gene can be broken down into 300 billion genome, the silicon circuit and the guide antenna can be parsed into a paired digital structure. These digital structures are randomly divided into half, allowing them to combine new design, or in combination with the next generation. Although the results are not always as good as those, you can screen out those unmetive results.
Lorne recalls a small episode. He and his colleagues have accidentally removed a large number of reactive parts in the circuit. Lorne puts it ratio as degraded organs. It is like a human appendix, once played a period of function, but later degraded.
Evolutionary engineering breakthrough tradition
Next year, NASA Space Technology No. 5 plans to make a first time an antenna assembled an antenna design with an evolutionary program. This antenna is made in the Amo Aviation Research Center of NASA. This antenna looks like a plain paper clip. It does not have a cubic inches. It has 5 sharp corners and a mitigation corner. The entire design took only 10 hours. It is completed by 35 Linux servers, and the ingredients of artificial intervention are minimal.
Jason Loron, from the Amos Research Center Evolution System Panel, said: "We only give our software to our software to enter the minimum antenna knowledge so that it can design it as you think the most suitable antenna." His This scene has broken the difference between traditional engineering and evolutionary engineering. Traditional engineers favor stringent and detailed drawings, they just use computers to make upset calculations and experiments. However, the evolution of the designer has traded out the entire design process, handed over to the computer and the design product itself.
The principle is still unknown
Even if these self-evolution is generated under the designated rules, it is still a difficult thing to analyze their principles. Bird and Lai Zel pointed out in their oscillator test research report: We have nothing to know how these circuits work. The mystery is the transistor. If you don't have a transistor, that circuit cannot be self-on the new line. Lorne found that the electronic engineering community is now divided into two factions. The first place advocates the principle of exploration. As long as you can get the expected results through the evolutionary process, you don't have to make it clear.
Another party is opposite. They wondered what happened, but unfortunately, researchers could not force this. Lorne is more willing to stand in the camp of the former, because this makes him feel easy. He describes his antenna design as "dark box art." This makes him and the followers of the lathequitry of Maxwell after the 19th-century British classic electric mechanics kicked away, and the latter is more preferred to perform accurate and textbooks.
Of course, this does not mean that there is no chance to reveal the dawn discovery in this field, and maybe a new discovery allows scientists to explain the inspiration of self-evolution of hardware.
Another interesting thing is that researchers have made a mistake in the test. They connected the circuit with inferior wire. In this case, the repair metal line means that a large number of valuable circuits have been destroyed, and the evidence that the circuit has gradually adapted to the harsh environment, and the evolutionary circuit does not want to see the metal line is repaired. Evolution is very embarrassed, it will use everything you handed over.
