traction control algorithms
Michal Brzozowski
I'm trying to gather some information about traction control systems,
in particular ABS and ESC. I'm mainly interested in what the inputs
are to those systems, and what algorithms the controllers use. I
understand they work by applying brake torque to chosen wheels.
I'm making a simple car simulation using Pacejka's model and want to
simulate those systems and see how they work. So I'm not really
interested in the engineering side.
I'm sure there are papers or websites that talk about this, any links
would be very welcome.
Thanks!
Michal
rcgldr
> in particular ABS and ESC.
Traction control usually cut's fuel to cylinders in a cyclical pattern,
and in some street cars it may just retard the spark. Stability control
uses individual wheel braking.
Michal Brzozowski
You are right. Replace traction control with stability control
everywhere in my post.
Michal Brzozowski
wrote:
> Hi,
>
> I'm trying to gather some information about traction control systems,
> in particular ABS and ESC. I'm mainly interested in what the inputs
> are to those systems, and what algorithms the controllers use. I
> understand they work by applying brake torque to chosen wheels.
Another question. Does anyone have Pacejka data for one tire model on
2 different sufraces? Like on dry and wet asphalt. Or any idea how to
modify the parameters of one tire to change the surface to more
slippery.
Thanks
Cronos
> Traction control usually cut's fuel to cylinders in a cyclical pattern,
> and in some street cars it may just retard the spark. Stability control
> uses individual wheel braking.
My old car was an Oldsmobile Alero and it had traction control (ETS) and
the manual said it used individual pressure on the wheels for traction
control which is the opposite of what you are saying.
rcgldr
Note the qualifier 'usually' regarding cutting engine power for traction
control. In the case of race cars, the rules dictate that traction control
is done via the engine, usually via the ECU. In the Indy Racing League,
the drivers can dynamically adjust the traction control on board.
Only a few race series, like the Dutch Supercar Challenge, allow stability
control (DSC allows any assists, and just classifies cars based on power
to weight ratios).
http://en.wikipedia.org/wiki/Electronic_stability_control
Ruud van Gaal
> I'm trying to gather some information about traction control systems,
> in particular ABS and ESC. I'm mainly interested in what the inputs
> are to those systems, and what algorithms the controllers use. I
> understand they work by applying brake torque to chosen wheels.
Racer uses:
traction_control
{
; Allowed front/rear wheel speed ratio (1.1=10% difference allowed)
max_velocity_ratio=1.1
; When to turn TC off
min_velocity_ratio=1.0
min_velocity=0
}
abs
{
; Anti-lock braking system
; Allowed front/rear wheel speed ratio
max_velocity_ratio=1.1
; When to turn TC off
min_velocity_ratio=1.0
min_velocity=0.1
; Braking factor when ABS is applied
braking_factor=1.0
}
Where a velocity ratio (left vs right wheel) is detected. These days
cars have so many sensors that engineers can use multiple inputs to
devise more intelligent controllers (mostly flowing around the car on
the CAN bus).
> I'm making a simple car simulation using Pacejka's model and want to
> simulate those systems and see how they work. So I'm not really
> interested in the engineering side.
Well, the engineering side is often a good way to start thinking about
implementations. ;-)
Cheers,
Ruud
Ruud van Gaal
> On Dec 12, 4:40 pm, Michal Brzozowski <rusolis.buzi...@gmail.com>
> wrote:
>> Hi,
> 2 different sufraces? Like on dry and wet asphalt. Or any idea how to
> modify the parameters of one tire to change the surface to more
> slippery.
Reduce D. For MF5.x Pacejka, you have the PDX and PDY variables.
From a piece of MF5.2 source:
Cx=pcx1*lcx;
mux=(pdx1+pdx2*dfz)*(1.0f-pdx3*gamma_x*gamma_x)*lmux; //
Different in Pac2006
Ex=(pex1+pex2*dfz+pex3*dfz*dfz)*(1.0f-pex4*sign_kx)*lex;
// Limiter on Ex (eq 23)
if(Ex>1.0f)Ex=1.0f;
Dx=mux*Fzn; // *zheta[1];
Kx=Fzn*(pkx1+pkx2*dfz)*expf(pkx3*dfz)*lkx; // K=BCD
(=stiffness)
Anything that reduces Dx also reduces grip. Pacejka targets more to
tweak the lambda values, lmux for example, to do your friction tweaking
(set it to 0.8 for example for wet). The nice thing about lambda values
is that you don't quickly miss effects in other parts, such as combined
slip etc.
For older Pacejka versions also look to decrease D (a2/b2).
Note that wet tires are quite ok these days, so don't go changing D to
half of what it was.
Cheers,
Ruud
Michal Brzozowski
> Michal Brzozowski wrote:
> > I'm trying to gather some information about traction control systems,
> > in particular ABS and ESC. I'm mainly interested in what the inputs
> > are to those systems, and what algorithms the controllers use. I
> > understand they work by applying brake torque to chosen wheels.
>
> Racer uses:
>
> Where a velocity ratio (left vs right wheel) is detected. These days
> cars have so many sensors that engineers can use multiple inputs to
> devise more intelligent controllers (mostly flowing around the car on
> the CAN bus).
Thanks Ruud!
I'm interested in how racer handles the increase in slip in both ABS
and traction control.
* Do you completely turn off the brakes/throttle when you detect loss
of grip, or set it to a smaller value, like 0.3 of the original
torque?
* What are the frequencies that you increase/decrease the torque with?
Wikipedia says that real life ABS do it up to 20Hz. I've gotten best
results with 50Hz for ABS, and much smaller, like 5Hz for traction
control. I'm wondering if this has any real life meaning or if it's
just a side effect of slip ratio oscillations at small speeds and the
like.
Ruud van Gaal
> On Dec 15, 11:56 pm, Ruud van Gaal <r...@racer.nl> wrote:
>> Michal Brzozowski wrote:
>>> I'm trying to gather some information about traction control systems,
>>> in particular ABS and ESC.
...
> I'm interested in how racer handles the increase in slip in both ABS
> and traction control.
>
> * Do you completely turn off the brakes/throttle when you detect loss
> of grip, or set it to a smaller value, like 0.3 of the original
> torque?
For ABS & TC, I turn off brakes & throttle completely. It seems that for
some cars, TC is achieved by skipping ignition for a few steps.
I'm sure ABS in real life is more complicated, with parts that are
actually breakable, but setting brakes to 0 seems ok.
> * What are the frequencies that you increase/decrease the torque with?
> Wikipedia says that real life ABS do it up to 20Hz.
I really has ABS & TC done in the innerloop, at 1000Hz thus. Probably
the resulting turning on/off of brakes / throttle is much lower, since
the wheel velocity ratio is looked at all the time, and these don't
change that quickly. I've never done any testing to the resulting
frequencies. I hope to get a Fanatec wheel where driving the brake pedal
actuator might invoke the need for a better ABS implementation (slower?).
Cheers,
Ruud
rcgldr
> for a few steps.
That would end up putting unburnt fuel and air into the exhaust, which
would ignite and probably do some damage, or at least create a lot
of unwanted head in the exhaust system. It's more likely that TC
systems skip fuel delivery to cylinders in a cyclic pattern perhaps
from 1 cylinder every 4 revolutions to 2 cylinders every 2 revolutions
switching which cylinders are skipped to keep temps even. The other
method is to retard the spark a bit, but generally that doesn't cut
much power.
Byron Forbes
news:d6JXm.28321$_b5....@newsfe22.iad...
Went to an Aussie V8 Supercar race a few weeks ago - you should hear
those things "BOOM" between gear changes due to ignition cut. Awesome at
first, annoying after about an hour.