Creating steering that feels alive 1

[Pcar928]

Cars such as the original Lotus Elan, the Ferrari F40, the early non-assisted NSX come to mind when I think steering feedback. The sort of feedback that has the steering wheel wrigling and writhing with nuances of info from the tires and road. The more I look into the subject the more it's made to seem like a black art to come up with that sort of steering feel. Just copy another car that has characteristics you like and hope for the best.

I have looked at alignment specs and such but you can't really draw conclusions soley on that. I understand the effects that pneumatic trail and the various offsets in the steering axis has in a static case, however are there other aspects to this sort of feel that are dynamic? Say from shocks transfering loads around (since no road is perfectly smooth) and changing the steered wheels P trail? Or maybe the force vector of the road against the tire acting around the steering axis varying direction slightly with oscillations of slip angle? Both of which can be connected so maybe a combination. Or maybe something else entirely?
I would just like to hear what some of those in the know have to say about it.

 

[evelrod]

Being a Lotus freak, as I am, and having spent a lot of time on them...One of the biggest problems I have encountered over the years is "tires"...Specifically, the WRONG tires can make all CABC's work a waste!

Now if you want "feed back" --- come take a spin in my Mini Cooper!

Rod

 

[Pcar928]

Sorry, CABC?

 

[evelrod]

Sorry my error...ACBC...Anthony Colin Bruce Chapman.
I should proof my posts, but I often forget.

Rod

 

[GregLocock]

Yes to most of your suggestions, add in friction and compliance effects as well. The design of the rack is crucially important, reverse efficiency and so on.

But yeah, a large proportion is tires.



Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Pcar928]

Is there any common characteristic that gives this feel of the tires that are fitted to these cars or are these cars simply designed around a certain arbitrary tire?

 

[Tmoose]

http://www.photoshoptalent.com/imag...eel/fullsize/steering wheel_48cc93e5123ab.jpg

 

[NormPeterson]

Add to the list damping created by the (almost universal) fitment of steering power assist, and I suspect that FMVSS limitations on force requirements may indirectly contribute as well.


Norm

 

[GregLocock]

But in the region where feel is important, efforts aren't very high. Admittedly getting a nice transition between the two is tricky.

OP- Yes tuning the steering feel and tire development go hand in hand.



Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Porsche356supersport]

I've been to some steering conferences recently where papers have been presented by ergonomists (human factor engineers) directly employed by OEMs to try and define the best haptic feeling for the steering system. The problem with this is that you try to define the feeling for a specific percentile group based on a fixed constant pattern of usage. The studies fail to equate the effects of varying physiology, psychological or varying environmental conditions. And in the case of a road car these are significant parameters that suffer drastic variation. There is also the issue of Mu to consider (the co-efficient of road friction). Consider the type of feeling you would expect on a snow covered road compared to driving on a smooth, low grip tarmac surface. It's interesting to consider that for some performance cars (BMW Z4, Mitsubishi EVO and Subaru Impreza) that HPS (hydraulic power steering) was chosen over EPAS (electric-PAS) because the OEMs felt the feel of the steering could be better tuned. Hope this was somewhat helpful.

 

[GregLocock]

Has anyone ever driven an EPAS system that had good steering feel?

I'm inclined to think that BMW's system (epicyclic in the column) offers the best chance of good feel, yet I've been told that this is not borne out in practice.



Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Porsche356supersport]

I would suggest that the BMW 5/7 Series and Volkswagen (Passat/Golf IV) have the best steering feel for an EPAS system. This is due to the configuration of the steering module which is similiar for both companies. The particular supplier uses a different set-up to other EPAS suppliers. I also believe that Ford of Europe with the Focus and Mondeo have done some excellent dynamic work on their EPAS system.

 

[TCAir]

Contact Jesse Black Ph.D. student at Clemson University. I was on a research project with him and we did all of this. I would give more info but I do not have my notes available. Truth is the OEMs know very little about "feel." We did synthesize it accurately in a drive sim we built. There are more variables than you can imagine.

Engineering has always been my love, but it ended up being my second career...

 

[Pcar928]

The EPAS in the late 90's nsx was superb as well, not as good as the non-assisted early models, but surprisingly engaging and communicative.

 

[cibachrome]

I dispute the statement that OEMS know little about steering feel and road feel. I happen to know several "companies" (1 OEM) that have enough technology to synthesis road feel to a level beyond even the BMW class IF they wanted to (or were allowed to pursue it. By that, I mean if they could get funding for the parts, an audience for the market, and a profit from the investment.

There are several almost insurmountable obstacles to getting this done. BTW: Hurrah for the use of the term Haptics, because this science is the key to a feeling objective in vehicle control (and not just road vehicles).

First, it is determined that only a fraction of vehicle evaluators are skilled enough to sense a road feel candidate. This is not an issue of training, but a skill brought to the table by a person who developed the skill at an early age (usually before 12 years old). Like certain eye reaction skills, you have it or you don't. This layer is a problem in vehicle development if the managing director does not have it (or worse the marketting and financial side of the business doesn't as a group, too).

Secondly, A vehicle in the hands of a proper evaluator must have steering and body controls "Que"d for an exemplary bandwidth. This is often misconstrued by controls people because its absolute bandwidth, not relative bandwidth. This means that the DC coherence and gain (via cross correlation of 4 signals), must be just as adequate as the 12 Hz bound. Too often, friction and damping become sacrificed for cost and quality reasons: Either the vehicle is good at low mileage and deteriorates at high mileage, or the other way around.

Thirdly, The presence of "prime" roadway signal (certain traits of transient aligning moment) can not be understated. A "good" set of tires is not enough with perfect body and steering control components. There are very important traits of a "Great" road feel tire. These factors are multiplicative: Any one sacrificed trait lowers the performance. One can not be used to augment another. The use of steady state properties to produce a candidate development tire can send you down a less than optimal path (so to speak) for road feel.

Forthly, A road feel evaluator can be tested on laboratory machinery to ascertain their evaluation skill ability. Its a visual as well as a haptic (feeling) response. Thus, machinery to produce various levels of multiply correlated movement and touch sensations is already in existence. (One in particular is in Canada: its a desireability for flying, too, you know).

Fifth (I'll take a pint, thankyou), it should be expected that a high powered, high frequency steer controller can do better than manual steering because the compromises made to maneuver at all speeds are too degrading of good or great road feel vehicle systems. You need some wattage for this.

Sixth(ly?) It is too often found that race car teams with multiple drivers can not use identical chassis and tire setups because the drivers do not bring the same feeling skill levels to the track. This becomes maddenly obvious when using one car amongst several drivers during testing sessions.

Lastly, a good experiment to run requires a roadway paved with a graduated level of lateral coarsness and a driver whose vision is restricted to eliminate path visualization.
This recognizes the ability of most good track racers to read the roadway as it evolves during a race. The chassis and tires read it, the driver reacts to it.

Oops, one more: The current problem with roadway simulators for roadfeel simulation and study, is the unacceptable screen delay of the visual field generator. Sure, an average "displacement" control driver will says it seems real. A "feel" driver will tell you its not even close. Many simulators showing F1 skits are driven by novices who wouldn't recognize the difference between an F1 and a Buick anyways. They just take the creator's word for it that its a Ferrari. Some of us also know that the Buick is wrong, too.

That's my spewage for the week...

 

[GregLocock]

Oh, I think you are overreacting to one BS post, that posted no salient facts, by someone who has probably only ever evaluated steering on Playstation style simulators. Ah here we go

http://ieeexplore.ieee.org/Xplore/l...282134/4282135/04282763.pdf&authDecision=-203

I don't quite see why somone who works on Playstation simulators for a motorsports degree is acclaimed as the expert on steering feel, perhaps TCAir could enlighten us rather than insult us?

But nice rant anyway. I'd add that we have to calibrate systems that can account for production variation. For instance, if you take 20 nominally identical torsion bar systems and measure the boost curves you quickly realise that while the general shape of the curve of boost force vs SWT is meaningful, any funny little humps and dips that you put in (especially for friction compensation) will not be meaningfully present in most of your production population. Now add in the changes in upstream and dowsntream friction between different cars, and as they age. etc



Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Pcar928]

Cibachrome
"Thirdly, The presence of "prime" roadway signal (certain traits of transient aligning moment) can not be understated. A "good" set of tires is not enough with perfect body and steering control components. There are very important traits of a "Great" road feel tire. These factors are multiplicative: Any one sacrificed trait lowers the performance. One can not be used to augment another. The use of steady state properties to produce a candidate development tire can send you down a less than optimal path (so to speak) for road feel. "

What are some of these "important traits" that make a great road feel tire?

 

[NormPeterson]

I don't quite see why somone who works on Playstation simulators for a motorsports degree is acclaimed as the expert on steering feel, perhaps TCAir could enlighten us rather than insult us?

I'll settle for a good explanation for why I can hardly keep the car in any arcade simulation on the road at any speed, but can go half a dozen autocross events in a row, of several runs each, without hitting a single cone. I know it's got something to do with "feel" and feedback.


Norm

 

[JohnLear]

Norm,
My experience with racing car games is that the 'driver' and the 'car' exist in slightly but disconcertingly different time zones. When the 'driver' makes an input there is a significant time delay before that input is seen on the scree

All those electronic calculations don't occur instantaneously, adding up to a lag which is enough to make the 'car' difficult to drive when you first start learning the game (and you crash a lot). With practice you subconciously adapt to the inbuilt delay and get somewhat faster with more control.

However, when you get substantially 'better' at the game (so you think) and the precise timing of 'events' gets more and more critical to improving your 'lap time', I think the subconcious tries very hard to make the inputs and outputs meet in real time and you then start to go backwards. At this point you start crashing a lot, again.

This delay is IMO the reason why even very good racing simulations are ultimately unsatisfying...

 

[GregLocock]

You may just be subconciously reacting to the poor physics of the games.

The physics model, in even the most sophisticated simulators, is a heavily streamlined version of the real thing. Typically arcade style games use 3 dof (dof=machine state variables) and a whole bunch of fudge factors, while professional simulators use maybe 25 dof, and a lot of lookup tables. These approximations are necessary in order to get the thing to run in real time.

To get the sort of results that I work with my models have hundreds of dof, and elaborate tire models that still bear (in many respects) only the crudest resemblance to the real thing, and likewise our steering hydraulic model is a very linearised model, same for the shocks. From that model I can accurately predict steering feel and handling, up to about 0.4g, and handling up to the limit with a great deal of extra effort. The downside is that the model will only run at say 1% of real time speed, in batch mode, ie not wasting cpu running the graphics.

The other issue of course is that in a game you only get limited feedback. On the other hand racing drivers claim to get some benefit from them.



Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.