THSCC Forums

Yeah, Donnie and I were having a side conversation about that. Now that I have my feet wet with my MaxQ, I'll have it "portable for use in Greenville.

Im sure your G data is MUCH more accurate than mine.

_________________
02 Focus SVT
STF 9

FYI I intend to have my GEEZ in my CM car on brand new slicks.

Given the fact that my GEEZ may read a little high combined with the g capability of my car/tires I should have some peaks over 1.5 g IF the weather is warm/dry and the rusty old driver is up to the challenge.

If all goes well I'll try to post a representative plot.

Then you can see why I cannot even imagine going back to even an R tire class, let alone a "street" tire class OR to a car where the driver's butt is more than 2 inches above the pavement.

Dick

_________________
Dick Rasmussen

FS 50 2018 Mustang GT

I *started* with that type of vehicle and I still don't get it. The same thing with the argument for R's versus streets. The challenges are where you find them...

Maybe I'm just easily amused???

Jim

Methinks you might be surprised at the current crop of DOT R's. Some Prepared drivers even run them now on asphalt.

_________________
02 Focus SVT
STF 9

OK, let’s check out the first part of the course. If you followed any of the RallyX write up, you’ll remember that it focused mostly on combining the best bits of multiple runs through a course to find the “perfect” run. This turned out to be a great approach since Carl and I had different strengths and weaknesses during our runs and they overlapped very nicely. I made very little effort to figure out how to improve the “best bits” because, frankly, I’m not entirely sure what good “dirt data” looks like yet. I have some ideas now and hopefully we can all learn something as the RallyX season progresses.

In this case, since I was the only driver, not only do I have less data to work with, I also tend to have the same strength and weaknesses every time. Compounding the issue even further, I was having car issues all day which caused the engine to sputter slightly making some of the acceleration curves a little wonky. To top it off, I forgot to record my second run. Not exactly the cleanest dataset to work from! Regardless, we’ll focus on the data from my best lap in more or less a vacuum and see if there is anything interesting. I will bring the other runs into play when we hit the optional slalom but, as you’ll see, it’s perhaps not as cut and dried as one would like.

Let’s look at the course map again but this time I’ll add a few new markers. The marker “opt in” represents the entry to the optional slalom and it is defined to be roughly where the apex cone was placed for the last offset. The marker “opt out 1” is defined to be near the end of the acceleration zone coming out of the optional slalom just as you were heading into what I called “the dirty slalom”. I defined an additional marker “opt out 2” which is just past the first slalom cone in the dirty slalom but let’s ignore that for now.

Course map with new markers and sectors:


We’ll start with “Sector 1” which is defined to be between “start of run” and “opt in”. I created a new track map and this time I brought it into Paint and showed off my mad skills. I placed orange dots roughly where the cones would be and it really illustrates a few key points and limitations of the currently technology.

Sector 1 with “cones” drawn in:


Seem familiar and yet a little off at the same time? What is it? GPS drift? Scaling issues? Bad accelerometer data? Think about it for a bit.

The next installment: The gospel, half-truths and damn lies!

Jim

My guess would be you combined the GPS data with the accellerometer data to generate the course map rather than pure GPS data. The software claims that is a more representative "true" path, but ? Could it also have been the DL1 was mounted a little cattywumpus which squewed the Acc data?
Yes, I've only read enough of the software help files to be dangerous...

_________________
SPIN or WIN!
there's no glory for going slow.

Jim,

Challenges in any class. I got hooked on autocrossing running tires and suspensions that wouldn't keep up with the average minivan today on all season tires.

ST tires are probably better than the first R tires I ran (Yoko A001-R's)

My comment had to do with being HOOKED on the G forces of a race car on hot slicks. I did not intend to degrade any class or driver. Sorry for not being clear. My bad.

Dick

_________________
Dick Rasmussen

FS 50 2018 Mustang GT

No misunderstanding Dick and I definitely didn't feel like you were degrading anything. I was commenting on the fact that I "don't get" the feeling of being hooked on higher g-forces, that's all. It's something I hear all the time-- race tires (in this case it's always R's and not true slicks) are like crack, I can never go back, blah blah.

But hey, there are lots of things I "don't get": Coffee, anything involving Will Ferrell and Duke Basketball among others...

I find it interesting more than anything as the grip is all relative to me and once I adjust to whatever I'm driving, it's all good fun no matter how high or low the limits are. Well, I take that back, I could do without it ever being as slick as it was in the morning of the first OCS RallyX.

Jim

So what is wrong with the course map?

The answer is a little bit of everything. In this case, while I had the DL-1 mounted reasonably solid, it still had a *touch* of wiggle room due to the tape. It did have a very minimal effect on the accelerometer data collected but I’d consider it negligible for our purposes. In general, I’d consider that data “the gospel”.

It seems that the overall perception of these tools is that they can instantly provide you with useful feedback. While that is true to some degree, they certainly have their limitations with regard to GPS capabilities and it usually isn’t apparent until you really start playing with one. The first limitation is simply one of sampling rate. Most (all?) of the sub-$1000 products only produce GPS data 5 times a second which while useful for general mapping, they can easily miss some course subtleties. The current generation of the DL-1 samples GPS data at 10Hz but that still isn’t enough to catch everything either. The bottom line? The course can wind up being a bunch of “half-truths”. It goes mostly where you think it does but you need to watch out for really quick inputs that may throw things off. Well, that’s not so bad, is it? Not so much. It is still very useful but that leads us to GPS “gotcha” #2.

And this is a biggie! The GPS data we are collecting is only accurate to a few meters! Wait, a few meters? Yep! Think about how big that is for a second in auto-x terms. One meter is roughly 3.3 feet so that means that if a “few” is defined as 3, that is nearly 10 feet! Most of the manufactures claim accuracy to 2 meters under what I would have to imagine would be very good conditions but that would still be only good to roughly 7 feet. In a sport where we are clipping the bases of cones, it’s clear the course maps aren’t going to help us pick a perfect line or figure out if we were even close to any cones. If you listen closely, you can probably hear the sound of bubbles bursting all over the Triangle. This sucks, right?

Wait, it gets worse! Not only is the GPS data not really that accurate but it “drifts” while you are sitting still! I’ve observed a much as a 15 foot “drift” while sitting stationary during testing. Cool, huh? Even cooler is the fact that the GPS still drifts while you are moving so your line is technically slowly wobbling all over the place as you drive. It’s kinda like giving a drunk a compass and a measuring tape and then asking him to survey a race track! Don’t believe me? Take a look at this next image.

Zoomed-in image of the area around the start line:


The green line is my first run of the day, the red line is my third run of the day and the black line is my last run of the day. I was *very* careful about lining up in the same place each time and yet it looks like I was all over the place. According to the scale in the upper left corner of the image, the first and last runs are nearly 8 meters apart! That’s more drift action than the last FnF movie!!!

OK, so this is totally useless, right? I mean, the course has to be a bunch of damn lies at this point, doesn’t it? Well, it turns out that through the miracle of cloning, it’s actually not that baaaaaad! Either that, or I’m getting ready to pull the wool over your eyes!

Next installment: Flocking to the truth.

Jim

Wow, did that last installment take an ugly turn or what? We saw evidence that the GPS data is totally worthless and I’m sure everybody must have the same question at this point: “What was the deal with all the sheep puns at the end?”

I know, I know, I was thinking the same thing! I’m sure by the last one ewe guys probably wanted to strangle me!

So what is the deal? No, not the sheep jokes, what is the deal with the GPS issues? Can we make these problems go away or at least minimize their impact? There are different techniques that can be applied in each case but I think the true answer is simple-- fuzzy math, of course! Yep, I’ll be here all week…

Seriously, is this really all gloom and doom? Let’s revisit each issue and see what we can do. As far as the sampling rate issue, there isn’t much we are going to do about that except spend more money. And by more money, I mean a *lot* more. To double the GPS sampling rate (20Hz), think roughly 5 times the cost of the DL-1 and go from there. Thankfully it isn’t linear and there are 100Hz GPS products on the market but they are north of $20k for the “basic” units. The sweet spot seems to be 10Hz so until the technology gets a lot cheaper on the high end, there isn’t much to be had there.

OK, so how badly does the sampling rate hurt us? Not that badly, really. Sure, we will miss some *really* quick maneuvers but, in general, cars don’t change direction fast enough to miss most of the good stuff. I’d be curious to see the course maps generated from the 5Hz products to see if much detail is lost in the slaloms and such.

The 2nd issue, the fact that the GPS position data is only accurate to a few meters really isn’t a big deal at all. We really don’t care about our *exact* position on the surface of the earth, we only care our position relative to where we just were. The bright spot in all of this is that it turns out the GPS is fairly accurate at distinguishing relative positions so it can tell if you move a few meters from your original position. Subsequently, the GPS speed values are accurate to less than 1 mph under most conditions since they are calculated based on those relative positions. At any rate, a map drawn purely from the raw GPS position data is going to be reasonably accurate relative to itself regardless of its true position in the world.

What about the last issue of “wandering” position data even when we are stationary? How much does it wander and how fast? How can we possibly account for that? Since I’m all out of sheep jokes, that will have to wait until I get back from the vet. No, really! I do have to go to the vet! Asthmatic cat having a rough day…

Jim

GPS devices that are "sub-meter" accurate ain't cheap. How do I know? Well we have ONE here at work. It's cost was about 20K more than a DL-1. Also the fact that it's backpack-sized antenna doesn't really fit well on a car's roof makes it use in our application kind of difficult.

_________________
Vincent Keene
'06 Ford Mustang GT (track rat)
'15 Dodge Charger R/T (yeah, it's got a HEMI!)
'07 Ford Fusion SE (205,000 miles and counting)
'98 Chevy Z-24 (retired)
'93 Acura Integra (Team SWB 24HOL Car)

Jim, re: drift. Is what you are showing in the start blowup really drift, or do you have the software set to offset the paths since the offsets are in run order?
Now if they would incorporate WAAS even the exact position accuracy would improve 2-3X. Unfortunately Europe doesn't use WAAS so they would have to build a receiver just for the US market. I don't expect that my 100.00 "streets and trips" with GPS uses as high a sampling rate as the DL1 yet I can "see" when I change lanes on the expressway when zoomed in, and my 300.00 Garmin that is WAAS enabled is more sensitive than that, it moves when I walk a couple steps with it in my hands and even measures my walking speed.

_________________
SPIN or WIN!
there's no glory for going slow.

I have a Garmin handheld unit that is very consistant, or rather, accurate relative to itself. Good for seeing where you are relative to where you were, but not downloadable to the computer so I can see the routes and play with them! Frustrating.

_________________
~ RallyX VP 2009 ~
- - - - - - - - - - - - - - - - - - - - - - - -
'92 Prelude
'76 Celica

Ash, that's the reason I don't like the Garmin can't play "what if" with the route (besides it deciding to change the route along the way... ). But that's a subject for another topic.

_________________
SPIN or WIN!
there's no glory for going slow.

Between runs I just check to make sure that no birds got sucked into the hood scoop. Sometimes I touch the intercooler to see if it's hot.


(Yeah, I got nothin')

_________________
Beauty is in the eye of the beer holder.