View Single Post
  #20   Spotlight this post!  
Unread 02-02-2004, 01:51
Scooter's Avatar
Scooter Scooter is offline
Galactic Doughnut!!
#0418 (Purple Haze)
Team Role: Programmer
 
Join Date: Feb 2003
Location: Austin, Texas
Posts: 117
Scooter will become famous soon enoughScooter will become famous soon enough
Send a message via AIM to Scooter
Re: Inertial navigation systems

I have been working on a 3 axis accelrometer for a satelltie with the University of Texas, and I can offer a few tips for those that are looking to develop one:

- I reccomend doing it as a seperate circuit and then feeding it into the UART of the robot controller as a digital stream. That way you can get accurate timing, etc.

-Make absolutely sure that you filter the analog vcc line on your microcontroller (an inductor up to power and a cap down to ground), as it needs an extremely accurate 2.5v to take a comparison reading from, and if the line is noisy, it'll throw off your readings.

-Use 2 sets of acclerometers and take the average of the both of them. Thus any error is reduced significantly.

-Search the web for analog smoothing algortihms. There are some really good ones out there.

-Shielding the lines, etc can be helpful, but the biggest difference that I have seen is fildering the analog vcc line. After I did that, my system's accuracy jumped up quite alot.

-There is a good white paper on this site on using the noise from the system to gain additional resolution. It has some good tips...

More later...
Bill
__________________
"Those who think that they know everything are especially annoying to those of us who do."