Author Topic: accelerometer for four rotor stability  (Read 7787 times)

0 Members and 1 Guest are viewing this topic.

Offline frank26080115Topic starter

  • Supreme Robot
  • *****
  • Posts: 322
  • Helpful? 2
accelerometer for four rotor stability
« on: June 10, 2007, 02:25:45 PM »
Me and a friend want to work on a four rotor helicopter over the summer, these things needs to constantly measure it's angle in mid air and adjust pwm duty for each motor constantly so it does not flip over in midair.

I know of two kinds of sensors to use, gyros and accelerometers, accelerometers can measure gravity and thus can act as a gyro
Barometers needs to be very very very very sensitive to do this, infrared and sonar rangers can't go too high or outdoors, not practical at all

The problem is, say the wind accelerates the helicopter at 0.5 G to the left, the helicopter will think that it is tilted 45 degrees to the right, and thus it will speed up the right side motor and slow down the left side motor in an attempt to rebalance itself. This means it will accelerate EVEN MORE to the left. So the slightest movement in mid air might cause the helicopter to go crazy and crash

I can get sample accelerometers for free from ST Microelectronics and Freescale, and they don't cost too much to buy (under $10, and SOIC). Gyros on the other hand, costs $30 each and Analog Devices are not gonna give me any for free, also not solderable by hand (ball grid)

The question is, is there a way for a MCU to decide whether the helicopter is moving or tilted? How does the wii remotes do it so well?
mechanical swivel devices holding the accelerometers will not work

cool site

Offline hgordon

  • Expert Roboticist
  • Supreme Robot
  • *****
  • Posts: 373
  • Helpful? 7
    • Surveyor Robotics Journal
Re: accelerometer for four rotor stability
« Reply #1 on: June 10, 2007, 03:16:30 PM »
Accelerometers are good for measuring tilt angles of your airframe relative to gravity (technically, this is referred to as "pitch" and "roll").   Gyros are good for measuring rotation, such as how the airframe is changing heading (this is referred to as "yaw").  So at a minimum, you want 2 accelerometers for pitch and roll angles, and one gyro for yaw rate.  Additional gyros for pitch and roll rates are useful for stability, but not absolutely essential.

There's really no way to know if the airframe is moving relative to the ground without an external reference such as gps, as it could be tilted, but wind is holding it in place.  I don't think you're going to see wind forces on the order of 0.5G unless you are flying in a hurricane.  In any case, you can add a pitot tube barometer to measure relative wind speed.

This will give you an idea of how to compute wind forces - http://www.noahs-ark-anchors.com/content/wind.htm

« Last Edit: June 10, 2007, 03:18:42 PM by hgordon »
Surveyor Corporation
  www.surveyor.com

Offline frank26080115Topic starter

  • Supreme Robot
  • *****
  • Posts: 322
  • Helpful? 2
Re: accelerometer for four rotor stability
« Reply #2 on: June 10, 2007, 10:31:25 PM »
ohhhhhh i totally misunderstood gyros then, i visualized spinning tops at the thought of the word and thought they measure angle

0.5G, thats going from 0 m/s to 4.9 m/s in a second, i think i can visualize me flying in that much wind, i want to do something like what (harvard or mit, forgot who) did and put a cmu camera on there and have it track stuff

thanks

Offline Admin

  • Administrator
  • Supreme Robot
  • *****
  • Posts: 11,703
  • Helpful? 173
    • Society of Robots
Re: accelerometer for four rotor stability
« Reply #3 on: June 11, 2007, 07:04:59 AM »
I bought this 5 DOF IMU last week but havent got it in the mail yet . . .
http://www.sparkfun.com/commerce/product_info.php?products_id=741

For its weight and simplicity its probably your cheapest option.

It'll take a month or two to report on my findings . . .

Offline hgordon

  • Expert Roboticist
  • Supreme Robot
  • *****
  • Posts: 373
  • Helpful? 7
    • Surveyor Robotics Journal
Re: accelerometer for four rotor stability
« Reply #4 on: June 11, 2007, 08:00:16 AM »
Sparkfun also has some accelerometers and gyros with digital interfaces (SPI or I2C) in case you don't have a sufficiently accurate a/d converter on your microcontroller (10-bits is not enough).  The LIS3LV02DQ 3-axis accelerometer has gotten good reviews.  I ordered SPI gyros direct from Analog Devices, as I believe the ones offered by Sparkfun only have analog interfaces.  Jon Hylands has experimented with a lot of these sensors, and can probably make some good recommendations.

In any case, for quad rotor designs and electronics, you might want to check out OpenSourceQuadroCopter - http://www.uavp.de/index.php?lang=en
Surveyor Corporation
  www.surveyor.com

Offline JonHylands

  • Expert Roboticist
  • Supreme Robot
  • *****
  • Posts: 562
  • Helpful? 3
  • Robot Builder/ Software Developer
    • Jon's Place
Re: accelerometer for four rotor stability
« Reply #5 on: June 11, 2007, 09:00:43 AM »
Admin, that's the part (5 DOF IMU) I use on my Bioloid 6-axis IMU...

http://www.bioloid.info/tiki/tiki-index.php?page=6-Axis+Bus+IMU

- Jon

Offline frank26080115Topic starter

  • Supreme Robot
  • *****
  • Posts: 322
  • Helpful? 2
Re: accelerometer for four rotor stability
« Reply #6 on: June 11, 2007, 10:05:04 AM »
10 bits is not enough??
thats 1024, 180 degrees divided by 1024 steps is 0.176 degrees resolution
thats not enough? i figured 8 bit conversion was enough, 0.703 degrees resolution

Offline Admin

  • Administrator
  • Supreme Robot
  • *****
  • Posts: 11,703
  • Helpful? 173
    • Society of Robots
Re: accelerometer for four rotor stability
« Reply #7 on: June 11, 2007, 10:19:28 AM »
well it depends on the accuracy you need . . .

but an IMU wont give a rail to rail 0 to 5V . . . for example, a very common accelerometer will give between 1.33V to 1.99V . . . not much for 8 bit! of course you could amplify that for each degree of freedom if you want . . .

doing the math for 8 bit . . .
360/(.66/5*255) = 10.7 degrees resolution

and 10 bit (which I dont really trust to be accurate)
360/(.66/5*1024) = 2.7 degrees resolution

Offline frank26080115Topic starter

  • Supreme Robot
  • *****
  • Posts: 322
  • Helpful? 2
Re: accelerometer for four rotor stability
« Reply #8 on: June 11, 2007, 11:03:37 AM »
is 3 degrees resolution too jerky? and why did you use 360? don't accelerometers only do 180 before being basically inverted?

i don't wanna spend $100 just for the sensor

probably going for a coanda effect UFO now

Offline Admin

  • Administrator
  • Supreme Robot
  • *****
  • Posts: 11,703
  • Helpful? 173
    • Society of Robots
Re: accelerometer for four rotor stability
« Reply #9 on: June 11, 2007, 11:08:17 AM »
Quote
is 3 degrees resolution too jerky?
Probably not . . . sounds reasonable to me

Quote
and why did you use 360? don't accelerometers only do 180 before being basically inverted?
because I wasnt thinking . . . lol . . . ok corrected math:

8 bit . . .
180/(.66/5*255) = 5.4 degrees resolution <= more likely
and 10 bit
180/(.66/5*1024) = 1.3 degrees resolution

Offline hgordon

  • Expert Roboticist
  • Supreme Robot
  • *****
  • Posts: 373
  • Helpful? 7
    • Surveyor Robotics Journal
Re: accelerometer for four rotor stability
« Reply #10 on: June 11, 2007, 03:15:41 PM »
is 3 degrees resolution too jerky?
3-deg is definitely too jerky for a quad rotor - the vehicle is inherently unstable, and you'll need much tighter control.  Even with higher precision and a fast sample rate, the controls are a challenge.

Quote
probably going for a coanda effect UFO now
Interesting (but loud).  I started down the path of designing quad rotor flyer, but am going with 2 coaxial rotors instead.

Surveyor Corporation
  www.surveyor.com

Offline Admin

  • Administrator
  • Supreme Robot
  • *****
  • Posts: 11,703
  • Helpful? 173
    • Society of Robots
Re: accelerometer for four rotor stability
« Reply #11 on: June 11, 2007, 04:40:44 PM »
Quote
3-deg is definitely too jerky for a quad rotor - the vehicle is inherently unstable, and you'll need much tighter control.
ummmm Im sure you can run some type of filter on it . . . what would you say is acceptable accuracy?

Offline hgordon

  • Expert Roboticist
  • Supreme Robot
  • *****
  • Posts: 373
  • Helpful? 7
    • Surveyor Robotics Journal
Re: accelerometer for four rotor stability
« Reply #12 on: June 11, 2007, 06:29:30 PM »
ummmm Im sure you can run some type of filter on it . . . what would you say is acceptable accuracy?
For stable hover without lateral translation, I think you'd need better than 1 degree absolute accuracy.  In any case, you'll get some drift, so it's a matter of deciding what is acceptable.
Surveyor Corporation
  www.surveyor.com

Offline Admin

  • Administrator
  • Supreme Robot
  • *****
  • Posts: 11,703
  • Helpful? 173
    • Society of Robots
Re: accelerometer for four rotor stability
« Reply #13 on: June 11, 2007, 08:21:41 PM »
Quote
For stable hover without lateral translation
Oh, I see what you mean . . . itll be stable with perhaps 10+ degrees error, but the error will cause a problematic lateral drift . . .

 


Get Your Ad Here