3

I'm building a quadrocopter and would like to prevent drift using a down facing camera. To do this properly, I need to get the height of the drone from the ground. Obviously the ground can vary, so maybe something like radar would be an idea, but I was wondering... how accurate are altimeters (electronic) these days?

Kortuk
  • 13,362
  • 8
  • 60
  • 85
Bojangles
  • 996
  • 12
  • 18
  • 1
    Why do you need to know the height? Because the speed scenery moves by will vary by height? I think that could be compensated for by a feedback controller, though, couldn't it? It's just seeking to minimize the movement of the ground, not calculate speed from it. – endolith Dec 08 '10 at 22:15

2 Answers2

5

If you base your design loosely on the Parrot drone, they use a downward facing ultrasonic rangefinder. I'm not sure of the exact brand, but it looks like a PING or Devantech rangefinder.

Different model numbers vary in their range and accuracy. Most are in the 2-5 meter range and have an accuracy of about 2-5cm. Obviously, the more you pay for the sensor will determine how accurate the sensor is.

Both Acroname and Sparkfun have good selections of ultrasonic rangefinders.

You may be able to also use a different type of rangefinding technology. I have seen IR and laser rangefinders used in many robotics applications. For a quadrotor, the ultrasonic rangefinder seems to be the best in terms of cost and performance.

mjcarroll
  • 1,974
  • 10
  • 16
  • Brilliant! Thanks for the great answer; ultrasound looks to be the way to go. I will use an altimeter as well, so I can get a larger range of measurements (but less accurate ones obviously). – Bojangles Dec 08 '10 at 21:38
2

The answer effectively boils down to two questions:
"How Much do you want to spend"
"What are your size requirements?

Realistically, there are three way to measure altitude without looking at the ground below:

  • GPS - Not very precise (~3m), unless you know people in the military.
  • Inertial Measurement - Expensive, if you want decent precision. Theoretically could work anywhere.
  • Barometry - Requires an air column. Sparkfun has one that claims a resolution of 9 cm.
Connor Wolf
  • 31,938
  • 6
  • 77
  • 137
  • 3
    A barometric sensor with a 9cm resolution sounds too good to be true, and I think it should be taken with more than a pinch of salt. First, they're absolute pressure sensors, so they have to be recalibrated for height with changing air pressure (weather). Then this is _resolution_, which means the theoretical height (pressure) change represented by 1 LSB. That's not _accuracy_, however. That depends on a number of factors, like linearity, and noise. If your LSB is drowned by 5 bits of noise your effective resolution reduces to 3m. Which would still be very good by any standard. – stevenvh Dec 09 '10 at 12:56
  • 1
    GPS note: actually, there's a second way to get precise height out of GPS: sit still for a day or so. Not useful for the project at hand, but the technique is used by folks like the National Geodetic Survey in some of their equipment. – Michael Kohne Dec 09 '10 at 16:06
  • 1
    You should be able to get down to ~1m with a single-ended GPS setup, assuming a relatively decent sensor (I believe that the uBlox line is pretty good for the cost). If a higher degree of precision is necessary for GPS measurements, you can move to a differential GPS setup. This is VERY expensive, on the scale of hobby projects, but very accurate (sub-decimeter or centimeter accuracy, depending on algorithm used) – mjcarroll Dec 11 '10 at 19:35
  • Thank you all for your comments. I will definitely take these into account when designing the stability system. – Bojangles Dec 13 '10 at 08:33
  • The 3m figure is applicable for positional accuracy in latitude/longitude plane, but not for altitude. – mehfoos yacoob Oct 15 '15 at 07:06