Inertial Navigation System (INS)

Discussion in 'The Projects Forum' started by bradstormer, Oct 9, 2010.

  1. bradstormer

    Thread Starter Member

    Aug 6, 2010
    65
    1
    hi all,
    i have been tasked with the design and build of an INS for a car.
    i need to be able to output information regarding trip distance and direction. Also the use of a member of the 8051 micro controller family must be used.
    i have checked the likes of wikipedia for information but most sites display info regarding space and aircraft.
    any help regarding this would be greatly appreciated
     
  2. bertus

    Administrator

    Apr 5, 2008
    15,648
    2,346
  3. windoze killa

    AAC Fanatic!

    Feb 23, 2006
    605
    24
    How accurate do you want this to be?

    How much money have you got to throw at it?

    You can get some "reasonably cheap" accellerometers that could do the job. You can also get an IMU but these are a lot more expensive.

    Using an 8051 maybe a bit on the hard side. We use an IMU in one of our small aircraft and we use a CPCI power PC card to do most of the flying. You could use the 8051 just to do the calcs and conversion.

    Good luck.
     
  4. gootee

    Senior Member

    Apr 24, 2007
    447
    50
  5. John P

    AAC Fanatic!

    Oct 14, 2008
    1,634
    224
    Maybe this is a school project where you're being told what to do, but if this is commercial you're nuts to do it with inertial navigation. GPS systems are just so cheap and simple that I'd say you need a very good reason not to use them.
     
  6. bradstormer

    Thread Starter Member

    Aug 6, 2010
    65
    1
    thanks for the replies everyone,
    ya its my final project in the course i am doing. the idea is to be a back up if GPS for a period of an hour or so. the accuracy should be fairly good, and funding for the parts is coming from the people running the course, within reason anyway!
     
  7. windoze killa

    AAC Fanatic!

    Feb 23, 2006
    605
    24
    That would be the only reason you would use inertial nav. If your vehicle was to go underground, ie. in a mine or similar.

    The IMU we use costs about 5000GBP so as I said not cheap but very good and reliable (which it has to be). You can get cheaper ones so look around. There are a lot out there that have RS232 or 422 outputs. 422 is very good in a noisey environment.

    Have fun
     
  8. bradstormer

    Thread Starter Member

    Aug 6, 2010
    65
    1
    quick question - is there an I.C that will give a direction, like north south eas or west?
    its ok -found them! they are called solid state compass and work by using magnetic field sensors.
     
    Last edited: Oct 10, 2010
  9. windoze killa

    AAC Fanatic!

    Feb 23, 2006
    605
    24
    Technically thats not inertial but would help in your accuracy.
     
  10. bradstormer

    Thread Starter Member

    Aug 6, 2010
    65
    1
    if i used an accelerometer in conjungtion with it, could the device the be considered to be an inertial measuring device?
     
  11. windoze killa

    AAC Fanatic!

    Feb 23, 2006
    605
    24
    I would assume so. Accelerometers are inertial devices.
     
  12. gootee

    Senior Member

    Apr 24, 2007
    447
    50
    http://www.analog.com/en/mems/products/index.html

    http://www.analog.com/en/mems/imu/products/index.html
    http://www.analog.com/en/sensors/inertial-sensors/products/index.html

    They have lots of other chips (accelerometers, gyroscopes, angular rate sensors, etc), but this one looks nice:

    ADIS16405: High Precision Tri-Axis Gyroscope, Accelerometer, Magnetometer
    Product Page: http://www.analog.com/en/mems/imu/adis16405/products/product.html
    Datasheet: http://www.analog.com/static/imported-files/data_sheets/ADIS16400_16405.pdf
     
    Last edited: Oct 10, 2010
  13. bradstormer

    Thread Starter Member

    Aug 6, 2010
    65
    1
    ok got a general game plan for the project.
    i have a formula to calculate position from the accelerometer o/p
    it is to integrate the o/p twice, which can be done using two integrator op-amp ccts. apperently tho any error in the readings can get quiet high!
    anyone know a way of combating this?
     
  14. wayneh

    Expert

    Sep 9, 2010
    12,123
    3,048
    In general, to combat data with a low signal-to-noise, you simply have to take more samples to be sure where the mean is. With enough samples, you don't care how broad the bell curve is. In your plan, I think this means you'd need multiple accelerometers on each axis. Probably not what you want to hear. But I suspect just going from one to two would make a huge difference, as long as their "noise" is truly random and independent. If your noise is from design problems and not independent for each accelerometer, then adding more won't help.
     
  15. mike8675309

    New Member

    Sep 25, 2010
    7
    0
    In Minnesota the University has been working with State and City transportation organizations on a technology to support Buses driving along shoulders of highways at high speed. This to avoid the cost of adding additional lanes to keep buses on time.

    To do this they need to have very exact fixes on where the buses in relation to the roadway and obstacles along the roadway. This Bus 2.0 technology includes a multitude of sensors. In particular, they have something called their VBAS or Vehicle-based Augmentation System which according to them:
    Their research might be useful to the Original Poster and since it is a public university, the information should be fairly available.
    http://www.bus2.me.umn.edu/

    Some of this technology came from earlier studies done to increase snow plow efficiency and safety in low visibility situations.
    http://www.its.umn.edu/Research/IVfieldtest/index.html
     
  16. bradstormer

    Thread Starter Member

    Aug 6, 2010
    65
    1
    thanks for the help everyone,
    i will probably use three accelerometers in this project and am trying to decide on which on at the moment.
    it is my first time using them and am finding the whole world of accelerometers a bit overwhelming!
    so far i think i need a two axis accelerometer with a range of +- 5g. (to cover the a car to a max acceleration of about 170km/hour)
    after that i'm not too sure
    i can decide on the analog or digital output after i take a better look at the micro controllers in the 8051 family, its the other information on accelerometers that i'm not too sure of.
     
  17. tom66

    Senior Member

    May 9, 2009
    2,613
    214
    One of the problems with IMS is known as drift.

    Assume that your accelerometer has an error of +0.05% (that is each reading is read as 0.05% greater than it actually is - which would be very good for an accelerometer.) If you take 5 samples per second and go for 1 hour you will have a drift of +8048%, which is absolutely massive.
     
  18. bradstormer

    Thread Starter Member

    Aug 6, 2010
    65
    1
    ya i heard about that,
    i was told the other day by my project lecturer that i'm not tied down to the inertial aspect of the project (although the brief says specificaly to research design and build an INS) but it would be nice to use an inertial sensor rather than something that i strap onto a wheel.
    but if i could even use the sensor to just give direction (calibration would be required on start up of the INS each time, not much of a problem) i would use the other sensor for distance, a magnet attached to the wheel for instance.
    i would like to have at least one inertial aspect though (i didnt realise the error would be so big!!!)
     
  19. windoze killa

    AAC Fanatic!

    Feb 23, 2006
    605
    24
    If you were to use a magnatometer or electronic compass in conjunction with accelerometers you should be able to counter the drift (in theory). Problem is these devices also have errors of accuracy. If these errors are in the same direction then it would compound the issue. Of course if this "device" is only to compensate for short losses of GPS then when you get GPS back you can correct the errors. Also if these errors are consistent you can then add compensation to your algos.
     
  20. gootee

    Senior Member

    Apr 24, 2007
    447
    50
    Have you read any of the application notes at analog.com?

    Um, by the way, 170 km/hr is the scalar magnitude of a velocity vector, not an acceleration.
     
Loading...