Professor Lee over the past five years to participate in a series of projects, including the guidance of a competitive super puppy soccer team to participate in the California desert at the challenging road racing, as well as to teach a dog backflip. As he displayed as the last one to do up so easy. He placed three feet high performers on stage with a rubber pad, puppy pad for backflips. The dog successfully performing a backward somersault, but on the sets of the edge, shrank to the ground. Lee said: "They are sometimes not so smart."
Professor Lee's research object is not a real dog, but the Sony Aibo (Aibo) robot. His research, some element of whimsical, but he was involved in the project - such as the guidance of the robot soccer team how to fight side by side, or let the autopilot equipped with sensors and small-engine car in heavy traffic city streets - will help him realize his own ambitions. In fact, that is simple, he just want to know how to make robots think and act like humans.
Professor Li cooperation organized by Penn Engineering and the Wharton School MBA program in a lecture entitled "Intelligent Robot recently: the next step will be how? (Smart the Robots: What's Next?)," The seminar said, let the robot engaged in a variety of tasks, like humans, also need to explore uncharted territory. For years, researchers have only focused on developing the basic technology of the robot work. However, Lee pointed out that although this technology has been progress, but human intelligence - even the ability to recognize faces - is still our current capacity can not match. He also for their own areas of research, a key question: "Why is the establishment of artificial intelligence will be so difficult?"
To replace human soldiers
The answer to this question will exceed the scope of academic research. In the U.S., the Pentagon has spent a lot - some analysts believe that over the years of its investment has reached $ 100 billion - Development of robots able to assist or replace human soldiers. U.S. Department of Defense's Future Combat System (Future Combat Systems, FCS) modernization program has been provided for the technology development funding for research. Li, where the Penn team won the Army Research Laboratory (Army Research Lab) $ 22 million in funding for the manufacture of independent operations in the theater of the robot.
Lee believes that although the United States, the commercial development of automated robots for military purposes, but the smart robot still has a huge business opportunity. He pointed out that the intelligent robots of Japanese studies in order to help the rapid expansion of the older age groups to do the housework.
In-depth study of the artificial intelligence aspects of commercial development may make the robot from one expert put it recently, "3D uses dull, dirty or dangerous (dull, dirty or dangerous) three purposes. March 31, Honda demonstrated a helmet-like device, it can read brain waves and transmit to Honda made a humanoid robot. With this device you can make, called "Love" (Asimo) robot to perform simple tasks, including moving its arm.
Lee said: "In Japan, robots as a service." And allow the robot to perform business tasks is relatively slow progress in the United States. "In the U.S., the National Science Foundation (National Science Foundation) has also funded many research projects, the United States for the commercial use of robot-related research, more money from the defense budget."
Lee envisages a future commercial application of robots. Of these technologies is likely to affect the future of industry and enterprise. He said: "We can create a lot of built-in sensors, actuators, and computing the micro-electronic technology, installations and equipment; The only problem is how to make them work smart and active in various environments. Future devices will not only store information, allowing users to play games, but also to provide more help. "
Professor Lee received the professional training of physics, holds a BA from Harvard University and Massachusetts Institute of Technology condensed matter physics doctorate. Joined Bell Labs in the 1990s, robotics had a strong interest. He wanted to know why - despite the computer technologies such rapid development - the robot is still unable to perform the task of humans and other animals can be easily completed. Lee is now studying biology, hoping to learn more about human calculation information in order to apply it to the world of robots.
Lee talking about the status of robot development, said: "These devices have been configured computer components and sensors, but people do not know how to use it." He believes that to create a smarter machine, the best The approach is to emulate the human brain.
The analogy of biological principles to promote technology development is not a new concept. He pointed out that the People used to think that as long as the wings will be able to fly, until the early 20th century, the aerodynamic concept the first time been applied. "At that time someone would wear a pair of wings, jumped from the downstairs - the outcome can be imagined, this does not work." The reason why the Wright brothers in 1903 flight, because they carefully studied the mechanics of bird flight behind principles. Birds vibration of wings in order to maintain balance and move forward, but they can fly in the air at full speed, the curvature of their wings to the body under pressure is greater than the pressure of the head, lift flight. "So the pilot's main job is to use vibration to the aircraft wing parts radians.
But the human brain does not like the bird's wing as easy to replicate.
Lee pointed out that the computer can use the "exhaustive" analysis to defeat chess champion, but it He said that the human brain is a very sophisticated instrument, far beyond the scope of the people to understand. Traditional computer algorithms rely on fast search and brute-force calculation can not make the machines become smart, so we need to develop a new algorithm to deal with these issues in different ways, so that the robot can work in a complex environment. we have to do some productive work, but if you want a machine to become smart, and dogs still have a long way to go. "
The robots have mastered the skills of some biological - Lee's Sony has demonstrated that love treasure robot. First, he would only electronic dog cried out loudly: "Hey," electronic dog turned his head at him. Then he walked to the other and saying: "hey, here", the dog also head turned to him. At the nanoscale in order to develop this robot technology, the designer must develop a sound sensor can simulate the human voice processing: the sound waves reach the left or right sensor speed difference can you tell the robot sound sources and head turning.
Lee said, with a gradual increase in mission capability of the robot to sense its surroundings and implementation of mobile capacity - will gradually enhance their artificial intelligence - without continuous human guidance will be able to perform certain tasks. This is why four Sony love treasure robot composition football team's participation in international football match is not just an interesting entertainment, but Lee and his students also become a valuable learning tool.
"UPennalyzers" - by Professor Lee, the other teachers and students of the Penn Academy of Engineering to create the football team - 4 built-in computer of the robot. They can use the sensor analysis of 3 × 5-m arena, and through wireless communication with teammates - but do not need human control in 20 minutes of the game.
Lee said the robot beginning of the game is cruel. Seen six or seven-year-old children play soccer? Everyone at the same time chasing a ball robot competition, so they either football trapped in the corner as long as half an hour, or put the ball into his own goal "But Lee play video UPennalyzers complex from the center of a recent game, show that the wisdom of the robot has gradually improve, to higher level of competition. "Now they can communicate: 'You are responsible for the eye on the ball, my defense'."
The team used the robot rather than humanoid robot is extremely difficult because the two legs to move the robot, In recent years, Li and his team from the Pentagon's Defense Advanced Research Projects Agency (DARPA) to obtain at least $ 3.5 million in funding, training robots to walk and manufacture a way to use the machine to walk independently by the sensors in difficult terrain people. The military advantages of such a device - in a dangerous war zone transporting heavy loads through the mountains or the desert - is obvious, but walking on two legs are still like the holy grail of the robot programmer. He said: "They want a weight of 200 pounds through the high mountains of the machine, because the car is unable to climb the high mountains."
New use of the lighter?
However, the wisdom of the development of unmanned vehicle - the project also received huge amounts of money in the DARPA goal is to achieve unmanned by 2015 one third of the U.S. Army vehicles can - Lee and his students the success and pride. Recently, they have short-listed the 2007 DARPA Urban Challenge, one of the six final teams to California Victorville an abandoned U.S. Air Force base entry.
"Small" - Ben Franklin team of Penn and Lehigh University students and teachers to give them the modified Toyota Prius take the name - in the city streets on their own driving, while complying with traffic regulations and instructions and to avoid collision with the other finalists. To make the race more realistic and challenging, the 50 test driver wearing a crash helmet driving. After watching the bystanders stood thick concrete barriers.
Carnegie Mellon Institute and General Motors co-won the tournament and 2 million cash award; Stanford University received federal funding, for the manufacture of the intelligent machine car. Due to insufficient funds, the Ben Franklin team wisely configured with fewer sensors, creating a power consumption less car - burning less gas. Lee said: "With the Prius, we will be able to use lighter provide the impetus for all the computers in the car."
Lee told the Penn audience, when the robot to grasp the dish into the dishwasher this highly complex human task, we know that the robot has reached a higher first-class wisdom. He said: the development of industrial automation robots spent us a long time, and now we need a way to let the robot learn how to live in the dish into the dishwasher, because every night, dishes and glasses the number of different. "
No comments:
Post a Comment