Why do quadrotors have four rotors?why not just one, like a helicopter?
You might think that, you know, a helicopter has one rotor, so what could be more simple than that? But the helicopter actually is more complicated mechanically.When a propeller turns, it generates a lift which is perpendicular to the blades of the propeller. This is not unlike a conventional propeller airplane or helicopter. But imagine opening a door. If the doorknob were mounted at the hinge, I would not be able to open the door. The hinge has to be offset from the doorknob in order for me to open the door. Similarly, if I have an object and I have a rotor in the middle of the object, then all that rotor can do is produce a force along the middle axis of the object. But if the rotor is offset to one side, then not only can it pull or push in one direction, but it can also have a net effect to turn. By regulating the speed of the four propellers, I can produce lift at any moment I want.
So how do you steer one of these things?
You?re only able to control propeller blade RPM, and you alter those to achieve motion. If these four speeds are identical, then the force will be identical and the lift will be identical across all of the rotors. Therefore the robot will just hover. Or, if I increase these speeds, it will accelerate along the [central] axis. To move it to the right, what I have to do is to make the left blade rotate faster than the blade on the right. When that happens, the lift generated by the blade on the left is more than the lift generated by the blade on the right. This causes the entire frame to tilt to the right. Now the rotors are no longer vertical, they?re pointing to the right. And once they?re pointing to the right I can use my old trick?I spin them all at the same speed and they?ll accelerate in the right direction. The robots in the music video are prototypes of a new-and-improved quadrotor. How are these ones different from previous models?
Over the years we?ve gotten a good sense of what you really need to have in this integrated system to get it to work. The new quadrotor has onboard sensors that are customized for high-speed applications. It has a radio that allows you to communicate with it on two different frequencies. It has motor controllers that control motor RPMs very precisely. And most importantly, it was built at a scale that we thought was most appropriate for indoor operations. The largest one that we?ve used indoors is about 1 meter (39 inches or so) in diameter, and the smallest one we built ourselves is about 8 inches in diameter. By scaling things down, we ended up improving the agility of the quadrotor and it?s actually much more robust now. How does the quadrotor make its own decisions?
There?s a processor onboard that does everything. It has to decide, or somebody has to tell it, what its destination is. It also has to figure out where it is, how to go from Point A to B, and then it has to determine how to spin the rotors to follow this path from point A to point B. It?s not unlike when you?re sitting in a car and you want to drive from, say, New York to Philly. You have a map, and then you make the decisions on how to steer the car to follow the roads. Everything below that is being done by the car, but this robot has to do everything.To do this, it uses accelerometers and sensors that detect angular rotational velocities. For outdoor operations it also has a barometer that can sense changes in altitude. There are also radios, which the robots use to communicate with each other through a way-station.
How much of that decision making is programmed, and how much is spontaneous? If I threw a baseball at a quadrotor, could it dodge it?
If you threw a baseball it, it probably could dodge it. If you throw it at 90 miles per hour that might be a different story, but it depends on the dynamics. So if it has time to detect the baseball, it can dodge it. It can react to surprises. How did you make the robots play the James Bond theme song?
There was no modification to the robots, but as you can imagine, the instruments were designed for human fingers. So what we had to do was actually customize the instruments. Because it calculates motion 600 times a second and gets feedback 100 times a second, this robot is actually more precise than you or I could ever be. The algorithms tell it exactly where to land and when. If you have to hit the note C so many beats after you hit F, that?s not magic?that?s prescribed by the music. But they also have to coordinate with each other, right?
When you play piano, your left hand and right hand are synced. Your brain basically has a clock, so that the right hand knows that 0.3 seconds after I hit this key, I need to hit that one. And the right hand knows not to hit keys that the left hand is playing, so the hands do not collide. All that logic has to be built in. But the biggest challenge was adapting the instruments. Getting the robots to fly very quickly in precise directions?that actually is a lot harder than getting robots to play the keys. How do you get them to work together?
The most important thing is sensing position. And the second most important thing is communicating that position to neighbors. The robots actually communicate with an overhead camera that we just bought off the shelf. They talk to the cameras and the cameras tell them where they each are 100 times a second. Then they have to talk to a way-station to find out where their neighbors are. So if you know where you are, and you know where your neighbors are, then you can figure out how to go from point A to B without bumping into your neighbor. What else have you used the quadrotors for?
We really want to use it in post-disaster damage assessment, such as mapping radiation levels after a nuclear disaster, or going and finding victims. I have this dream that the first responders to 911 calls will not be law enforcement personnel but robots. Robots can put eyes and ears on the scene much faster than you can with policemen or women. What they cannot do yet is take the right action, but often knowing what is happening and where it?s happening is the most important thing, and that you can do in a very cost-effective way with robots. They can also be useful in terrorist-type situations?you could send a robot in to get information. Have you tested the quadrotors in an emergency situation?
A We?ve already done some experiments in the Philadelphia area and in Japan near Fukushima. At Sendai, a town about 100 kilometers from Fukushima, we did some experiments in a collapsed building, where we were able to send a larger quadrotor to areas where you could not walk because debris would block your way. We were able to construct multifloor maps of the building without any human intervention. Normally, in the presence of radiation, communication links fail. But with autonomous robots, you don?t need communications. It can go and build a map, and then come back and tell you what it looks like. What are the limitations of a quadrotor?
The small size makes it difficult to pick up large objects. If you put it in a 30-knot wind, it?s going to get blown away. How are quadrotors different from something like a UAV?
Quadrotors are completely autonomous, and I challenge any human being to produce those kinds of maneuvers using remote control. You can?t do it. You need the extraordinary precision of algorithms? a human being just can?t do it. What are you planning to do next?
Our goal is to push the envelope in terms of autonomy, in terms of the number of vehicles that can collaborate with each other, and the number of sensors that the vehicles can carry. knowshon moreno knowshon moreno sovereign citizen komen ohio state basketball chrome for android hatchet
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.