Someone said that would be hard or impossible to do?
I agree it'd be useful. Possibly more useful would be a g-g diagram. Those aren't too hard to keep an eye on while driving the track. A couple years ago I tried that. It'd be great for replays but it's hard to watch everything while driving hard. The center circle is a g-g diagram. The other circles are the traction circles (true circles in this case but elipses would be easy to do too). The blue bars indicate slip ratio. The green lines in the tire circles show the slip velocity vector at each tire (slip angle indicators, basically).
http://performancesimulations.com/files/ToddSim-GenIV-10.wmv
Take opinions on this sort of thing from people that have never done it with a grain of salt. Generally speaking, folks usually don't know what they're talking about
Last edited by jtw62074, .
Blackbird, do you have any experience or interest in artificial neural networks (ANNs)?
One thought I kicked around for awhile a couple years ago sounds a bit similar to what you're looking at doing and might very well have some hope of working with an ANN somewhat well. The reference points would have to be in pairs evenly spaced around the whole track, probably. I.e., on the left and right track edge every xx meters would be two points across from each other at about 90 degrees to the track direction. If you used the relative position (compared to the car position) of the nearest __ pairs of points in front of the car (always the same number of points) and fed them into some input nodes of an ANN, then used other things like velocity, slip angles, yaw torque, steering angle, etc., an ANN might very well learn how to spit out a throttle/brake and steering position or gain rate. If you leave "training on" the cars can accidentally figure out how to get around the track faster periodically while running the sim.
You could train the ANN by driving a few laps yourself until they can at least manage to get around on their own, then let them go and play around to get faster. Or if you have a lot of time, use a particle swarm algorithm and let the AI try to learn its way around the track all on its own. If I were to do it I'd start out on a small oval track or something simple.
I haven't tried that, but did experiment a year or so ago with an ANN and particle swarm algo to drive a basic steering controller for Virtual RC Racing. This was much simpler as it was trying to follow a predetermined line and that's all. So it was basically using only one node at any given time. The ANN did indeed learn most of the time how to whiz around the track at a pretty good pace, but it was not quite competitive with a hand written and tuned controller. However, I only played with this for a couple of days before being forced to drop it, so I'm pretty sure it could have been improved a lot. It's really just evolving neural network controllers via something similar to natural selection, but it's a simpler approach to implement than using genetic algorithms. I'm not sure if it would have been easier or harder to do with regular cars versus the little RC cars though. There was an F1 type of game awhile back that did this so well that the cars were virtually unbeatable. They had to use "younger" generations in order to keep it fun for the masses.
During my brief attempts at this in VRC, one funny thing was there were some tracks where the car would usually turn around immediately at the start and learn the track backwards instead of forwards, requiring the addition of learning penalties in a desperate attempt to get the cars to go the right way around!
Usually after 5-10 minutes or so (time accelerated so "simulated time" is much higher) the cars learned to get around the track fairly quickly. It's fun to watch them learn and get better without touching anything for sure.
You might find this is all a lot more difficult to do than you're anticipating. You don't know yet what troubles you're going to run into that will soon be eating up all of your time. So may I make a suggestion? Try using slip angles, rpm, position, and whatever else you can get directly at first. If you're happy with the results you could then expand this into using sound to determine rpm and understeer/oversteer and so forth to emulate human perception. My worry is that if you do this right away, you're adding a complex layer on top of everything right away which has a high chance of seriously clouding issues and slowing you down. For example, if the car is always turning the wrong way at the corner, is it because of a miscalculation in the reference point stuff or is it something funky in this "human perception" layer? When doing experimental stuff like this I usually start simple and go from there. With this approach I get a much greater insight into what the system is really doing (compared to what I thought it would do) and am more likely to see where things are wrong or can be improved just by watching what the car is doing. You start to see a lot of things other people won't.
One thought I kicked around for awhile a couple years ago sounds a bit similar to what you're looking at doing and might very well have some hope of working with an ANN somewhat well. The reference points would have to be in pairs evenly spaced around the whole track, probably. I.e., on the left and right track edge every xx meters would be two points across from each other at about 90 degrees to the track direction. If you used the relative position (compared to the car position) of the nearest __ pairs of points in front of the car (always the same number of points) and fed them into some input nodes of an ANN, then used other things like velocity, slip angles, yaw torque, steering angle, etc., an ANN might very well learn how to spit out a throttle/brake and steering position or gain rate. If you leave "training on" the cars can accidentally figure out how to get around the track faster periodically while running the sim.
You could train the ANN by driving a few laps yourself until they can at least manage to get around on their own, then let them go and play around to get faster. Or if you have a lot of time, use a particle swarm algorithm and let the AI try to learn its way around the track all on its own. If I were to do it I'd start out on a small oval track or something simple.
I haven't tried that, but did experiment a year or so ago with an ANN and particle swarm algo to drive a basic steering controller for Virtual RC Racing. This was much simpler as it was trying to follow a predetermined line and that's all. So it was basically using only one node at any given time. The ANN did indeed learn most of the time how to whiz around the track at a pretty good pace, but it was not quite competitive with a hand written and tuned controller. However, I only played with this for a couple of days before being forced to drop it, so I'm pretty sure it could have been improved a lot. It's really just evolving neural network controllers via something similar to natural selection, but it's a simpler approach to implement than using genetic algorithms. I'm not sure if it would have been easier or harder to do with regular cars versus the little RC cars though. There was an F1 type of game awhile back that did this so well that the cars were virtually unbeatable. They had to use "younger" generations in order to keep it fun for the masses.
During my brief attempts at this in VRC, one funny thing was there were some tracks where the car would usually turn around immediately at the start and learn the track backwards instead of forwards, requiring the addition of learning penalties in a desperate attempt to get the cars to go the right way around!
Usually after 5-10 minutes or so (time accelerated so "simulated time" is much higher) the cars learned to get around the track fairly quickly. It's fun to watch them learn and get better without touching anything for sure.You might find this is all a lot more difficult to do than you're anticipating. You don't know yet what troubles you're going to run into that will soon be eating up all of your time. So may I make a suggestion? Try using slip angles, rpm, position, and whatever else you can get directly at first. If you're happy with the results you could then expand this into using sound to determine rpm and understeer/oversteer and so forth to emulate human perception. My worry is that if you do this right away, you're adding a complex layer on top of everything right away which has a high chance of seriously clouding issues and slowing you down. For example, if the car is always turning the wrong way at the corner, is it because of a miscalculation in the reference point stuff or is it something funky in this "human perception" layer? When doing experimental stuff like this I usually start simple and go from there. With this approach I get a much greater insight into what the system is really doing (compared to what I thought it would do) and am more likely to see where things are wrong or can be improved just by watching what the car is doing. You start to see a lot of things other people won't.
Last edited by jtw62074, .
Ok, figured it out. Will post here just in case others have problems.
I had installed the ATI drivers while I only had two of the video cables. Plugging in the third one then didn't work. So if you make that mistake too, just uninstall the drivers through the CD app that comes with it and reinstall. It'll reboot and all three monitors will work.
At least in my case it did. Now to go do some driving
I had installed the ATI drivers while I only had two of the video cables. Plugging in the third one then didn't work. So if you make that mistake too, just uninstall the drivers through the CD app that comes with it and reinstall. It'll reboot and all three monitors will work.
At least in my case it did. Now to go do some driving
Does anyone else have the HD5870 and a triple monitor set up? I got everything I thought I needed, but can't get more than two monitors to work. The monitors are Dell 24" ones with all different types of connectors. My understanding was that I could run two monitors through the two DVI ports while the third monitor must be through the DisplayPort. Got back a little while ago with a DisplayPort cable that goes straight from that monitor to the card, but it won't display more than two monitors at a time. In Catalyst Control Center all three displays show up, but the DisplayPort one is disabled. I can't "extend" the desktop to the third without being forced to disable one of the others.
The ATI's documentation that came with the card didn't even have directions for the HD5800 cards on the CD. Just a bunch of manuals for the older cards. :chair:
The ATI's documentation that came with the card didn't even have directions for the HD5800 cards on the CD. Just a bunch of manuals for the older cards. :chair:
Ahha. I get it.
Not sure what you mean by that.
It was pigs after all. First thing that popped up on a search was this:
http://www.futurepundit.com/archives/003971.html
I read about it on Yahoo's front page earlier this year. I didn't realize it was three years old already.
I read a Yahoo News article fairly recently about some research on this. I think they had done it with pigs or something. That's great if they're already testing this on humans. Thank goodness for animal testing
The 800x600 resolution is actually surprisingly good. It looks higher than that to me. Going below that though might be a problem because I'm not so sure that many (if any) games these days support such low resolutions. Even finding ones that would do 800x600 was a hit or miss thing. Most handled it, but not all.
It's the only computer thing I've ever had that has increased in value on the used market since buying it. Sad.
Maybe somebody can use those new flexible LCD screens to come up with something better and cheaper. I just spent $1600 on a 24" triple monitor setup with an RD 5870 card or whatever it is. I'd pay $500-$700 (again) for a beefed up Z800 with bigger screens or equivalent optics. The effect is quite staggering really. Just want bigger screens or the ability to pull them in a bit closer. Resolution was good enough already even at 800x600.
I'll take a new chance at growing teeth, please. Where do I get in line?
First thing I'd do is eat a steak.
Sorry. You have a "TV License" over there?
I must have misunderstood what you were getting at then. The handling characteristic I was talking about was off-throttle oversteer and pointed to the MR-2 as an example. You responded:
I took that to mean you thought the reason the car had a lot of off throttle oversteer was because the moment of inertia was low.
Yes, of course that's right. It has nothing to do with what I was talking about though.
I've thought about that too. I've got a Z800 virtual reality headset which is pretty awesome, but it leaves me wishing the images were bigger. Something with mirrors might be able to do a better job. I'm not sure, but I have a sneaking suspicion that might be how the really big virtual reality headsets with the big pod looking things on top might be working. Perhaps there might be reasonably big screens up top and you view them using a mirror system of some kind.
I'm anxious to see what you come up with and how well it works :-)
Fair enough. If they want to train people to drive lousy cars that don't drive like real race cars should then so be it
That's not right about the MR-2. The moment of inertia has nothing to do with understeer/oversteer balance. At the limit if there's a yaw torque into the turn it'll oversteer. That comes down to tire lateral force distribution between the front and rear and that's about it. All the moment of inertia controls is how large the yaw acceleration is given a certain yaw torque. A car with a huge moment of inertia that oversteers will still oversteer. It'll just accelerate around the yaw axis more slowly than a car with a low one (it'll spin out more slowly). Low moment of inertia does equal 'more twitchy' though and typically a design goal is to get that as low as possible so the car responds to the yaw torques quickly. That part's right.
Last edited by jtw62074, .
I just tried both again and the Skippy seems a lot better than it was before (default setup). Perhaps that's my imagination since it's been so many months since the last drive in it. Pretty sure I remember it being a lot worse than the Spec in this regard.
I tried both at the skidpad just now. At a constant 60mph when the throttle is dropped quickly they both spin to 90 degrees by the time they stop in just under three seconds. This was keeping about 90 degrees steering lock throughout the slide without countersteering at all. You're right. To me it looks like the Spec accelerates more quickly initially and is more sensitive to this. I remember the reverse being true by a long shot when the Spec racer was released. Am I remembering wrong or has the Skippy changed in the past six months or so? I was always at the front of the pack in the Spec racer, but would be off the pace by three seconds or so at Infineon just because I could barely control the car. I should give it another go and see if I move up at all.
Yes, that's true of course. My point was more that it's not really a design goal to make a car spin around when you let off the throttle like that. I understand that some cars do this (the first Toyota MR-2 springs to mind), but it's not something the engineers and drivers will generally want. You want the car just to go straight to the slip angle peaks and stay there if at all possible in all conditions. Some cars do this better than others of course.
I tried both at the skidpad just now. At a constant 60mph when the throttle is dropped quickly they both spin to 90 degrees by the time they stop in just under three seconds. This was keeping about 90 degrees steering lock throughout the slide without countersteering at all. You're right. To me it looks like the Spec accelerates more quickly initially and is more sensitive to this. I remember the reverse being true by a long shot when the Spec racer was released. Am I remembering wrong or has the Skippy changed in the past six months or so? I was always at the front of the pack in the Spec racer, but would be off the pace by three seconds or so at Infineon just because I could barely control the car. I should give it another go and see if I move up at all.
Yes, that's true of course. My point was more that it's not really a design goal to make a car spin around when you let off the throttle like that. I understand that some cars do this (the first Toyota MR-2 springs to mind), but it's not something the engineers and drivers will generally want. You want the car just to go straight to the slip angle peaks and stay there if at all possible in all conditions. Some cars do this better than others of course.
Last edited by jtw62074, .
I heard this about the Spec Racer Ford too. This had lift off oversteer too, but not nearly as bad as the Skip Barber car. I don't know if they updated that car too though. Just have to give it a try and see.
I'm fine with the real car doing this, but the amount just seems really over the top. Why someone would engineer a car to handle that way is beyond me. A bit of oversteer makes sense, but going to the point where it just completely spins in a fraction of a second is just a design flaw in my opinion. A friend of mine who's a race car engineer ran at a racing school at Laguna Seca. I think that's the Skip Barber school but could be wrong and am too lazy to check online right this second.
He pointed out that the cars there were set up really loose, but he didn't mention trailing throttle oversteer specifically. If it was his car to race he would most certainly change the car quite a bit towards making it more neutral.
It's just odd and something seems amiss. In my sim I have to try pretty hard to make a car drive that horribly when no throttle or brake is being used. A lot of people seem to really love it though. Just different taste I guess.
Another thing to point out with the Skip Barber car is that in reality you have people trail braking with those cars into the turns. That's just impossible with the iRacing version unless you go really far forward with the brake bias. Then once you let off the brake just enough it'll spin. I've not been able to set up the car to transition from full braking to no braking and full steering without spinning. There's an area in between there where you have to immediately jump from brake to lots of throttle or around it goes in a real hurry. Then once you're on the throttle it's all understeer. Seems odd to me.
I just can't get used to driving that thing. Maybe in reality I couldn't either
I'm fine with the real car doing this, but the amount just seems really over the top. Why someone would engineer a car to handle that way is beyond me. A bit of oversteer makes sense, but going to the point where it just completely spins in a fraction of a second is just a design flaw in my opinion. A friend of mine who's a race car engineer ran at a racing school at Laguna Seca. I think that's the Skip Barber school but could be wrong and am too lazy to check online right this second.
He pointed out that the cars there were set up really loose, but he didn't mention trailing throttle oversteer specifically. If it was his car to race he would most certainly change the car quite a bit towards making it more neutral. It's just odd and something seems amiss. In my sim I have to try pretty hard to make a car drive that horribly when no throttle or brake is being used. A lot of people seem to really love it though. Just different taste I guess.
Another thing to point out with the Skip Barber car is that in reality you have people trail braking with those cars into the turns. That's just impossible with the iRacing version unless you go really far forward with the brake bias. Then once you let off the brake just enough it'll spin. I've not been able to set up the car to transition from full braking to no braking and full steering without spinning. There's an area in between there where you have to immediately jump from brake to lots of throttle or around it goes in a real hurry. Then once you're on the throttle it's all understeer. Seems odd to me.
I just can't get used to driving that thing. Maybe in reality I couldn't either
Last edited by jtw62074, .
Tony West made my avatar the day he got banned. It's one of their West Racing teddies
I like the Solstice a lot, but hate the Skippy. It's just massive off-throttle oversteer. I can't see the point of setting up a car to do much more than tuck in the nose when you let out of the go-faster pedal. It's atrociously balanced if you ask me and I can't set it up to drive like a car ought to at all.
I have most of the iRacing cars. My favorite is probably the Radical once I get it set up more to my liking. The Solstice is great in that you can steer with the throttle and brake a lot more than with the other cars. I find myself using throttle and steering together in a much more fun way than with most of the cars. My other favorite is the F1 Lotus. Spending a lot of time in that. The IndyCar is good fun too, but so far I've only run that at the Indy Speedway (good fun) and Talledega (boring) aside from a few laps at Laguna Seca.
The Skippy's my least favorite by far, although it's been many months since I've run it. If they'd get rid of the horrible trailing throttle oversteer I'd enjoy it a lot more. Star Mazda is a lot more fun to me.
I'm not involved with KartSim any longer so can't say anything for sure other than they're working on it. I'll leave that up to them to discuss out here in forum-world though.
We've been hogging Tony West on Virtual RC Racing for many years now so his time is limited. Chris is still at it though.
The National Highway Transportation Safety Administration released data on nearly 400 vehicles in at least two SAE (Society of Automotive Engineers) papers dated 1992 and 1998. There are two versions of this data out. One in metric and one in SI units. This is the latest metric one with about 80 vehicles:
http://www.not2fast.com/chassis/sae1999-01-1336.pdf
I've referred to this in my little hobby sim test bed over the years many times. Do other sim developers? I don't know, but it'd be surprising if the iRacing guys didn't. If the exact vehicle you want isn't specified you can still get a pretty good estimate by looking at similar vehicles. At least you're unlikely to be off by 100% or something using this.
To answer one of your earlier questions: Yes, the three values for yaw/pitch/roll are indeed used in sims. More specifically you specify a 3x3 matrix called an "inertia tensor." The yaw/pitch/roll values go along the diagonal while the remaining values can just be 0. I.e., you wind up with an array for each car that looks something like this:
Car(0).InertiaTensor(%X,%X) = 100
Car(0).InertiaTensor(%X,%Y) = 0
Car(0).InertiaTensor(%X,%Z) = 0
Car(0).InertiaTensor(%Y,%X) = 0
Car(0).InertiaTensor(%Y,%Y) = 125
Car(0).InertiaTensor(%Y,%Z) = 0
Car(0).InertiaTensor(%Z,%X) = 0
Car(0).InertiaTensor(%Z,%Y) = 0
Car(0).InertiaTensor(%Z,%Z) = 150
In the rigid body dynamics code this is used to calculate the angular acceleration (in full 3-D) given all of the forces on the car. If the car is made up of multiple rigid bodies, each body gets its own inertia tensor.
In Virtual RC Racing we used this data straight out of the car designs in Pro-Engineer in which every single component has its own tensor. This meant the software calculated the cars' inertia tensor using all the masses and mass distributions of every single piece of each car. So it's possible to get very accurate if you have the original, complete designs in good engineering software without testing it directly. I doubt other sims have done this, really, but my point is it's possible to get this without testing the vehicle. A test is the best way to go about it though.
Last edited by jtw62074, .
This isn't a weight transfer effect. With a car that's understeer in a trimmed condition (constant speed skidpad test such as the first part of the video above) the lateral g is limited by the lateral force the front tires can produce. If they have enough to pull 0.9g while the rears can pull 1.05g (or 2 or 3 or 20g for that matter), the car will do 0.9g on a skidpad. The rear tire slip angles will only climb high enough to maintain that 0.9g. They have 'leftover force' yet to be tapped. If you kick the car sideways a bit to a larger slip angle, the rear tires will make more than that original '0.9g worth' of force.
Translation: Chuck the car into the turn a bit and you'll get more than 0.9g or whatever the skidpad rating is. It's something that can be (and frequently is) analyzed with a yaw moment versus slip angle diagram. This is not a weight transfer effect at all. You will see more weight transfer, yes, but that's because you're pulling more g rather than the other way around.
This is where suspension tuning for the track comes in. By increasing rear roll stiffness, the front '0.9g limit' increases because there is less weight transfer at the front. The rear 1.05g limit decreases. As long as the rear can still pull a bit more than the front, the car is still understeer. Voila, your skidpad numbers increase in a trimmed condition.
Last edited by jtw62074, .
Didn't realize it was so old. Was new to me and I got a kick out of it
Gymkhana spoof
Yeah, pretty scary
FGED GREDG RDFGDR GSFDG