See also TractionControlArm, TractionController

Hopefully one will clean it up, this info was just going down the drain...

* got almost all the high level logic stuff done

* starting to tie hardware + high level together

<blundar> I should probably have some more interesting stuff to post / share / etc. on 2004-09-27

* tractionboard = one of the "peripheral" ARM boards repsonsible for traction control through monitoring VR wheelsensors

* I went all out with the code

<blundar> did proper statistical tests

<blundar> i.e. analysis of variance

* has calibration procedure

* normalizes wheelspeeds to account for minor variances from VR sensors so you can even use different types of VR sensors and it will deal

* computes variance during calibration

* then it basically examines the wheelspeeds during normal operation

* uses the normalized values from calibration to normalize each sensor reading

* then does a weighted ANOVA

* and the weighting basically determines how sensitive it is

<chrisge> what is the reference for what wheel speed should be?

<blundar> reference is determined during calibration

* I'm figuring on having a button or something to push to signal calibration

* calibration procedure = drive straight with cruise control on or otherwise at a constant speed

* it measures 256 samples

* takes average, variance of each sensor

* to get an idea of how "noisy" they are

* then the ANOVA makes meaningful statistical comparisons of sample data vs calibration data

* the ARM has *plenty* of horsepower

* I figured doing proper statistics would be a good way to use it

----

What gear?

* Are you planing for use in every gear?

* it shouldn't matter which gear you're in

* for calibration at least

* It shouldn't know what gear you're in, should it?

* it shouldn't matter

* it's purely using wheelspeeds, 4 VR sensors

* slippage will be clearly manifested by 2 wheels fast than others

<chrisge> what if my front tires are not on the ground?:_)

<blundar> tilt sensors :)

* seriously, it has two, 2-axis tilt sensors

* you bring up a very good point tho - I hadn't thought of the RWD tires lift scenario yet

----

calibration

I was thinking about the calibration process. I believe it is possible to make it without an explicite button from the user:

When we collect the wheelspeed ratios of the 4 wheels to histograms, the peak of the histogram should be the no-spin - a good reference.

----

offtopic - no names - will be deleted

working for the nation's largest private provider of healthcare services to the military: I'm also strongly against military but I wouldn't feel anything bad about running computers for a hospital - even if it was for the troops.

"I think that any country has a right to protect whatever its interests may be" - yes, as long as it's not invading other countries ( we know what happened when Hitler decided to protect Germany's interests :-( ), or it does not commit terrorist acts (like the Nagasaki bomb: the Hiroshima bomb could be argued to be a justified act of war needed to save lives elsewhere by proving the power, the 2nd bomb was clearly the biggest terrorist act ever - appr. 30 WTC ). Anything regarding international affairs should be settled with the consent of EU - anything else endangers world safety rather than making the world a safer place.