Scientists at Stanford has designed an artificial intelligence system which permits robotic helicopters to educate themselves to fly complicated stunts by observing other helicopter performance that carry out the similar maneuvers. It can outcome in development of autonomous helicopter which can do a complete air show of difficult shunts on its own. It is one of the latest examples for newest technological inventions.
Team members involved:
This project is carried out under Professor Andrew Ng who performed the research with their graduate students named Morgan Quigley, Timothy Hunter, Adam Coates. The stunts carried out by such intelligent helicopters are far more complex than any other computer controlled helicopters. They have undergone various learning algorithms for these helicopters that assist them to learn by themselves by just watching other expertise and skilled helicopters.
This experiment was placed in a significant demonstration of apprenticeship learning in that robots learn by observing other expert helicopters performance as mentioned above. Stanford’s artificial intelligence system learned how to fly by observing the four foot long helicopters flown by specialist radio control pilot Garett Oku. The advanced helicopter can learn and carry out actions such as rolls, loops with pirouettes, traveling flips, a knife-edge, stall-turns with pirouettes, a slapper, an Immelmann, an inverted tail slide as well as a hurricane, denoted as a fast backward funnel.
Existing autonomous helicopter were capable to fly stunts by easily replaying the exact finger activities of an expert pilot making use of joy sticks present on the helicopter’s remote controller. But the major difficulty was the uncontrollable variables including gusting winds owing to which this is not very advance. To find solution to this problem, the researchers had decided a new concept that is Oku along with other pilot make the whole airshow routines fly and each movements of the helicopter was been recorded. As Oku repeated a maneuver many times, the route of the helicopter predictably varied slightly with every flight. At this note, the learning algorithms developed by Ng’s team were capable to determine the ideal route that the pilot was looking. Thus the autonomous helicopter learned to fly the routine better and more constantly.
Advanced intelligent structure:
This intelligent and advanced system contains some instrumentation accumulated on the helicopter and some on the ground. These instrumentations watch the direction, orientation, velocity, acceleration, position as well as spin of the helicopter in numerous dimensions. A ground based computer chews the data, create quick evaluations and beams new flight directions to the helicopter through radio 20 times per second. Some of the essential instrument it employs is Magnetometers, Gyroscopes and Accelerometers, but later earth’s magnetic field is used to figure out in which way the helicopter is pointed. The accurate location of the craft is monitored either by a GPS receiver on the helicopter or through cameras on the ground. These advance intelligent helicopters of a recent generation seems to be very reliable, very robust helicopter that can fly just as their human counterparts.