The online racing simulator
Searching in All forums
(635 results)
jtw62074
S2 licensed
I actually did post some stuff about NKPro, but changed my mind and deleted it. I'll move along now
jtw62074
S2 licensed
On VRC, indeed it's a real challenge with the keyboard and most people are really put off by it immediately without a proper controller. You couldn't run the real cars with a keyboard either, it'd be just as hard as in the game really.

On the real RC engineering level things are actually quite high tech. VRC was done in colllaboration with Serpent, one of the manufacturers (Pieter Bervoets, the founder of Serpent, is the owner of VRC), so I spent a lot of time talking with the engineers, Michael Salven and Rene Cornella (I stayed at Rene's place for about 6 weeks once). When Michael was designing their F-1 car (RC version of course), he had his CAD file open when I walked in to his office one day. I thought it was a real F-1 car on screen there. Nope, he had modelled the RC version after a real one and they were working on the prototype right then. Was very cool to see

Anyway, the vehicle dynamics of RC and big cars are the same really. It's just the numbers you plug into the model for weights and so forth that are different for the most part. The drivetrains are different of course though and there are some slight design differences, but all in all it's the same thing as a big car. Tires are similar in that force is a function of slip angle/ratio/load/etc. (just like any deformable body rolling on a hard surface), so again, you can use the same model there, but with different inputs.

Either way, I'm enjoying NkPro, and am also really looking forward to seeing the new tire improvements Scawen has come up with for LFS soon! These are sharp guys here on this LFS project and I'm enjoying watching them continue to improve this. Things can only get better and better, so let's just keep on supporting them. I think with all the time I've spent on LFS it's probably cost me only a few pennies per hour of entertainment. Well worth it!
Last edited by jtw62074, .
jtw62074
S2 licensed
Quote from Shotglass :for one its based on experience (havent doen any slides like that in a real car yet but ive done plenty with rc cars on all sorts of tyres) and with time you get a feel for what a car can do and what it cant possibly do
and its also that kunos mentions that you can feel the 3 stages of his tyre model (grip ... slipping at the edge ... and sliding) and somebody here voiced his concerns that it sounds like the model switches between those stages instead of a smoother more realistic transition which incidently is exactly what it feels like

I wouldn't worry too much about how the transition occurs. Stefano has had a lot of experience with tire models. I'd bet it transitions just fine. Isn't he using a Pacejka model?

Quote from Shotglass :
the problems begin with the way you enter that slide ... you step on the gas pedal and the tail starts to overtake the front ... so far so good ... but once the tails stars to step out the car instantly snaps (so much for not yawing quickly) to a full 90 and stays there
but it gets worse ... the usual reaction to this is lifting your foot from the loud pedal and countersteering ... the result of this is that the car instantly turns a 180° and you find yourself still sliding perpendicular to the track but facing the other way this time

Oh, ok. I haven't tried that in nKPro as I'm usually driving pretty nicely without a lot of sliding around. However, if something sudden happens right around 90 degrees that could indeed be a bug. I've had problems with that in the past too. Took awhile to figure out what was wrong and fix it though. I'll give it a try and see if it feels like mine did.

Granted, don't get too upset about that. As long as you're driving in the normal range or just a bit over that, a bug like I had really doesn't effect the handling at all. I.e., everything's just fine until you really get the back end waaaay out. It's a bit like low speed models where a car will float around a bit when it's supposed to be sitting still, or slide slowly down a hill even with the wheels locked. It doesn't effect the realism at all once you're going more than a foot per second or whatever (some very, very low speed). Lots of professional vehicle models used in engineering research have had this same problem. Many folks just ignore it because they don't really care what happens at anything less than typical driving or racing speeds.
jtw62074
S2 licensed
Quote from Tweaker :I don't see how an R/C car and real car tire can compare. Maybe Todd can explain that, but I find it very hard to believe they are the same. A pebble or dimple is about the equivalent of a pothole or tree log. And to have the car weigh in to about the weight of your floorpanel carpet, quite a lot differs... in huge ways.... especially if you have experience through 'feeling' the car though remote control :zombie:

Assuming a clean surface, of course: We've actually been testing foam RC tires on a machine for quite awhile now in order to find out what the lateral force vs. slip angle curves look like for different sizes and durometer readings, and indeed they're quite similar to full sized, air filled, rubber tires. The cornering stiffness is a lot lower (peaks are at very high slip angles), but the curves look the same overall. So really you can model these RC car tires with the same model you use for big cars. Just change a few parameters here and there and off you go.

Granted, this was not done on a carpet surface with electric tires, although we may do that in the future too. I expect them to look pretty similar. It's really just about finding what slip angle the peak is at as well as overall grip as a function of load so you have load sensitivity. All tires look pretty much the same in that regard, even the tiny little ones
jtw62074
S2 licensed
Quote from Tweaker :I think the point he was getting at was the fact that you can crank the steering to full lock, and there is actually hardly any understeer or even oversteer. It is so neutral and so stable. Driving the track normally, it is very possible to just do full-lock turns and you will be just fine.

Ok, I must have misunderstood. Still, this is isn't necessarily wrong. Really depends on the wheelbase/cg height ratio. If it's greater than a certain value then you'll start to get understeer at some point. If it's lower than that you can very well continue more towards oversteer as you increase steering lock (to a point of course, as you said, approaching 90 degrees isn't going to do it.)

As long as the tire data and model, plus the mechanical stuff on the car is right, the behavior should be correct. I haven't played much with the setup in nKPro yet so haven't really examined anything too closely.

My point really was that if you get a real car to some huge sliding angle, then neutralize the steering and release the throttle, it is entirely possible that the car will straighten up. You can really design a car to do more or less what you want it to do, so just because that behavior exists in one sim and not in another (of two different cars and sets of tires), doesn't mean one of them is wrong.

Quote from Tweaker :
If you crank your wheel at 90 degrees, of course you won't get oversteer from that, like on this skidpad, but I think it is very hard to create any oversteer with the little 1600 unless you have some uneven surface or just do some wild control behaviour. If you mount the rain tires (or whatever those RG tires are), try the car with those. To an extent, those feels kind of nice, and have a bit more possibility of sliding. They still have grip, but they are loose for pushing it faster (but those tires have lower grip levels anyways, so oh well).

EDIT: Doh, some words for understeer, I meant to say oversteer.

And if we really want to talk tires, maybe wait till this new LFS patch too.

I haven't tried switching tires yet. Will give that a go. Thanks for the tip.
jtw62074
S2 licensed
Quote from Shotglass :um ? no ... have you ever tried the f1600 on the skidpad ? try to creat oversteer ... you can hold that thing at full 90° with full throttle easily ... and it gets worse ... if you countersteer too much the car will suddenly snap back form a full 90 ... theres just no way any car would ever react like the nk cars in a slide

Why do so many people perpetually insist this behavior is impossible? If you have more rear grip than the front this is indeed what will happen provided you don't yaw the car so quickly it doesn't get a chance to straighten back up.

NkPro is using real tire data from Avon, which is pretty tough to beat. (That data is available publically at their web site, by the way, so anybody can use it). Seems to me Stefano got the combined lat/long force stuff down just right. Feels very good to me indeed Smile
jtw62074
S2 licensed
Hi Sebastien, long time no hear

The Novodex engine is very impressive to me in regards to collision detection/response. The constraint system for multiple contacts and joints and so forth is wickedly fast too. I don't get the claims many make about "fake physics" at all though. Seems rather unfounded.

Here's a neat video:

http://video.google.com/videoplay?docid=3103103511684048644
jtw62074
S2 licensed
Quote from Forbin :Plain and simple truth is Ageia fakes the physics in a way that LFS would never stoop to.

Errr... How so?
jtw62074
S2 licensed
Quote from Shotglass :a bit hard to tell the error without any scale ...

It's easy. Open up that pic in a paint program and count pixels like I did

Quote : but the thing the approximation is constantly significantly above the measurement so your simulated engine has a larger power band and higher peak power which should make the car quite a bit faster

Yes, but that's just one of the curves from when I was messing around with some new development stuff to make sure things shifted around correctly when things were changed. It wasn't meant to be spot on. I just plopped it together real quick to illustrate my point about it being unnecessary to do all the 3D mesh stuff and wait hours or days to produce a pretty good curve. As I said, there are several flow restriction values that are completely unknown through ports and passages. Garbage in, garbage out. In addition, I'm still working on it, so bear with me please

As said before, if some of the unknowns are tweaked it fits almost exactly, but that's not any sort of validation of a model. What counts is that you're pretty close when making reasonable estimations of the unknown stuff that has dramatic effects. If we want to be picky about it, the motor from the dyno test is a '97 engine that's a little bit different from the one I took measurements of.

Anyway, try this next one instead if that's what's important

http://www.performancesimulations.com/files/2stroke-C.JPG

If the tail at the high end throws you off, don't despair. We did not have data for the very top end of the curve or the bottom end before the clutch engaged.
Last edited by jtw62074, .
jtw62074
S2 licensed
Quote from Shotglass :hmmm you almost have me convinced there ... but those curves deviate a bit too much to really get me excited
also could you estimate about how much time it would take to come up with an engine simulation thats at least as accurate as yours ?

Interesting viewpoint. At peak power the error is only 4.5%, and that wider area past the peak is 10%. Below the peak torque the data was unknown and indeed should drop similarily to the sim curve. Average error is within 5%. That's with my wild guesses at flow rates at all the ports and so on. From an engineering standpoint that's considered excellent

Anyway, if I massage the flow rates and so forth it fits so closely you can barely see a difference, but that doesn't validate the model properly and to a science person it looks like obvious cheating to get the curves to fit

The amount of time it takes to write something like that completely depends on the individual. It took me several months and I've done it before a couple of times, so it's not a small task, unfortunately. A full blown 4-stroke model takes quite a bit longer to develop.
jtw62074
S2 licensed
I don't recall anyone doing any such thing
jtw62074
S2 licensed
One must decide whether one is going to do something or not at all before deciding where to fit it into her schedule. There's no harm in discussing things and dreaming up ideas in the mean time.
jtw62074
S2 licensed
Why don't you tell us what you REALLY think?
jtw62074
S2 licensed
There are different levels of abstraction you can aim for when writing an engine model. It's not necessary to write a full 3D FEM system in order to make a very realistic engine simulation that produces torque/power curves that are close enough to reality. I've written three or four engine models over the past decade or so. They range from simple to very complex, but they don't need to be 3D mesh type things that take forever to run to produce very good results. (They do, however, take a very long time to write if you've never done it before and are not already rather knowledgable about engines. My latest engine model is arguably more complicated than my vehicle dynamics system from a programming perspective.)

The drag racing simulation I wrote, Straightline Acceleration Simulator, has an engine simulation in it that I wrote mostly while in high school (QuickEngine Builder). If you check out the demo, you'll find it runs a complete torque/power curve in a fraction of a second. Granted, it's a much simpler model than my latest stuff, but it works quite well regardless and is plenty fast enough for use in games. Since then I've done a lot more complicated stuff that does everything you guys are talking about and more, but they do take more execution time to run.

For a good portion of the past few months I've been writing a 2-stroke engine simulation based on what I've learned writing 4-stroke models for inclusion in Virtual RC Racing 4.0. So far the results are quite good and we fully intend to have this engine tuning in VRC as it's very much a part of real racing.

The setup options that can be changed in the model are very extensive (all the port timing events and passage sizes and so forth, it's a full blown wave action simulation), but we're not going to have more than a handful of them available for the public to play with. I.e., in reality (in RC anyway) people play with different exhaust pipes, glow plugs, and tune the fuel mixture, and that's about it. They might grind the exhaust port a bit to change the exhaust timing to shift the power curve around a bit, but usually only the pros or very experienced racers mess with that. However, one of the cool benefits you get from it is that when weather conditions change the curves are effected realistically. You need to change the mixture a bit or maybe the glow plug in order to maximize yourself for that weather condition.

Exhaust pipe selection moves your power curve around, trading off low end torque for top end power or vice versa. In the current VRC version we allow a limited form of engine tuning so people can already play with the exhaust or select different engines, but it's all based on multiplier tables running off a real engine torque curve and so on rather than an actual simulation like I'm working on now. Still though, nobody complains about it in our community and simply sees engine setup as another thing to play with to get the car running how they want. In fact, quite a lot of them have asked for more engine variables to play with.

Attached is a dyno run from the simulation in its current state. There's still a fairly long way to go yet, but you can see how it compares to a real engine. The green lines are the torque/power curves of a real engine, except for the low end part where the torque curve is flat (this was at a low rpm where we didn't have any dyno data.) The red and black lines are horsepower and torque, respectively, produced by the simulation. As you can see it matches up quite well, especially for a model that runs the entire curve somewhere between 0.5 and 1.5 seconds or so. So you really don't need to have a super complex model that takes hours or days to run like the manufacturers might use in order to get pretty close results.

As for the various arguments pro/against engine simulations in racing sims, I must admit I find the "it'll complicate things and I don't know anything about engines" argument a bit ironic. I agree with others that you're already setting up the chassis, which is a very complex thing indeed to do, and that can easily make the difference between winning and losing a race. Most of you guys know quite a lot about setting up a car and I'd bet it's mostly because of the sims you've been playing. You learned to do it. And you can learn to set up an engine as well. In fact, you might find it easier to set up an engine than the chassis You really don't have to be sculpting your own combustion chamber and port designs in 3D or anything like that. Nobody does that but the engine manufacturers anyway.

Engine tuning isn't about getting stage 1 this, then stage 2 that, and so on, where whoever has stage 4 of everything blows away the competition. That's indeed very arcade and I suspect many of the folks that are against this idea are picturing that happening. I agree that if that was the case it would ruin LFS for a lot of people. However, if the setup options were realistic it might be quite fun and really not ruin things the way folks are thinking.

In my big car sim I regularily play with engine tuning by running curves off my QuickEngine Builder and find it to be really fun to do. What most people probably don't know is that making an engine produce absolutely as much horsepower as you can make it create is actually going to kill you. I.e., if your engine is going to blow up at 7000 rpm and you try to produce literally as much horsepower as possible, what you'll do is choose parts that get the torque to peak as close to 7000 rpm as possible. Then you can tell your friends about the 5.7L 600HP combination you came up with. Then you show up at the track and find you have absolutely no torque coming out of any of the corners and are getting blown away by guys with 400HP.

Yesterday I was running a 500HP car and swapped in a 750HP motor in my sim. My lap time on a minute and a half lap only improved by 2-3 seconds. Why? Because this particular track only had one straight long enough to take advantage of the power. You can't do a whole lot with the wheels spinning all the way down a short straight between two turns. Adding more top end power just makes it harder to control. So really, it's not necessarily a matter of "whoever makes the most power wins." You need to make power over a certain rpm range depending on your gearing and the track. Your numbers won't be as impressive as the next guy's, but he's left scratching his head as to why you've got 100HP less than he does and are smoking him

Anyway, if done right, I think it'd be fun. Plus, it might be fun to try different "engine manufacturers'" engines by downloading them. It might be fun and open up a new area for bragging rights
jtw62074
S2 licensed
Congratulations, Scawen! Best wishes!
jtw62074
S2 licensed
No way
jtw62074
S2 licensed
Colcob, your memory is right. Indeed that's what can happen in a tire model very easily. My own all suffered from that problem until only a few months ago when it suddenly dawned on me how to fix it (including the Virtual RC Racing model).

jtr, you still wind up with a traction circle, it's just that things won't really converge properly on it. Also, while driving well inside the traction circle, things need to be right too, otherwise the car can feel like it very suddenly loses grip at some point.
jtw62074
S2 licensed
<poof>
jtw62074
S2 licensed
Suit yourself

Edit:

Ok, I had to come back and amuse myself.

Here's what was said that got me into this discussion:

Tristan: (Force Dynamics comes closest, but it's only 1g sustained,

You: No. It can not exert sustained acceleration. To feel any kind of acceleration on yourself, you need to MOVE.

I said it wasn't necessary to move. Just pop your feet up in the air and you will feel exactly as though you were accelerating at 1g. I meant "forward," but didn't say so. I figured everyone would understand that, and they did, with the exception of yourself

You don't have to move at all to feel an acceleration. If you're stationary in a gravitational field you feel exactly as though you are accelerating. You are right now, "feeling an acceleration" straight upwards. Are you literally accelerating? Outside the context of 4D space where most of us spend our intellectual time, no. I agree. Your position isn't changing so you aren't accelerating. Nobody is arguing that.

However, you are indeed "feeling as though you were accelerating" at 1g. (Is it ok if people use the term "feel an acceleration" there? According to GR that's just fine because it really winds up working out just like that. You'll learn that later in school hopefully.) I.e., if there were no gravity and you were pushed upwards at 1g, it would feel exactly the same as if you were sitting on Earth, not accelerating at all. And in the latter case, there is no movement. So the statement you made in reference to a motion platform that sits at an angle to give the impression of forward/rearward/sideways sustained acceleration:

"It can not exert sustained acceleration."

Is sort of correct if taken literally word for word, although if you want to split hairs you don't "exert" an acceleration so the statement is meaningless. But I wasn't going to be anal and pick apart your sentence. I understood what you meant and just carried on.

Anyway, the next statement you made:

"To feel any kind of acceleration on yourself, you need to MOVE."

Is categorically false for the reasons given. Remember, the verb there is "feel." I.e., "to feel as though one is accelerating," not to literally be accelerating.

What's most amusing to me about this whole thread is that the people that know nothing about physics at all, or know less than you or I or Dr. V. over there might, understood this immediately.

"To accelerate you need to move"

Yes.

"To feel any kind of acceleration on yourself, you need to MOVE."

Nope. I am feeling exactly as though I were accelerating straight up at 1g in the absence of gravity right now and I'm not moving anywhere.

The tilting motion platforms are a snazzy idea because they can make you feel as though you were sustaining an acceleration of some kind in the direction of interest.
Last edited by jtw62074, .
jtw62074
S2 licensed
Amazing... The gentleman you're arguing with indeed has his Ph.D in physics (chaos theory) and is an author of the following papers. I believe the first one is his Ph.D thesis, but could be mistaken.:

"Faster than Lyapunov decays of classical Loschmidt echo"

"Classical Loschmidt echo in chaotic many-body systems"

Somehow I'm not surprised he currently teaches physics at a university in Slovenia. Ah well, perhaps you can teach him a few new things about gravitational acceleration versus gravitational force anyway while clearing up any misconceptions he might have about General Relativity illepall

Either way, have fun
Last edited by jtw62074, .
jtw62074
S2 licensed
Good grief, this thread went crazy.

My point was that you can not tell the difference between these two situations:

1) Stuck in a box strapped to a chair tilted back 90 degrees.

2) Being in 0 gravity while accelerating forward at 9.81 m/s^2.

Is gravity a force or acceleration? In the context of this discussion, who cares? The point is you can't discern between acceleration from being pinned to the Earth from an actual, literal acceleration in a direction straight "up." I think they taught us that in 9th grade. Enjoy your university studies

If you tilt the chair back 45 degrees you feel exactly the same as you would if you were horizontal in a gravity field of 70% Earth's strength while accelerating forward at 70% of 9.81 m/s^2. My point was you do not need movement to feel as though you are accelerating. You just need to be pressed up against some huge thing like the Earth to feel it.

And no, lying on your back with your feet straight up in the air will not feel like you're in a car accelerating at 1g because the "downward" component of the acceleration is now gone. I didn't say you would. However, you most certainly do feel exactly an acceleration of 1g "forward" in that you are facing upwards. You are not moving in this situation. Therefore, you do not require movement (an "actual" acceleration) to feel acceleration.

I can't believe I'm having this discussion and that point is being totally missed by someone so educated in physics.
Last edited by jtw62074, .
jtw62074
S2 licensed
You are feeling 1g acceleration in that case. And without any motion.
jtw62074
S2 licensed
You seemed to have it covered. I had nothing to add
jtw62074
S2 licensed
Lay on your back with your feet straight up in the air. 1g
jtw62074
S2 licensed
Geeze, Forbin, I thought you'd been around longer than that

http://www.performancesimulations.com/files/ToddSim11a.wmv <--750 HP at the flywheel, 2500 lb car. Watch the slip ratios. That's wheelspin, wheelspin, wheelspin on those short straights.
http://www.performancesimulations.com/files/ToddSim13a.wmv <--3500 lb car, 350HP, 0.95g skidpad, no downforce, and no problem either with or without diff preload
http://www.performancesimulations.com/files/ToddSim7a.wmv

Best really to go in here and right click/save as:

http://www.performancesimulations.com/files
Last edited by jtw62074, .
FGED GREDG RDFGDR GSFDG