The cost of stage notes.

Robin Fleguel Wrote:
-------------------------------------------------------
>
> Trust North Americans to 1)invent a system capable
> of guiding bombs into toilets from 26,000ft and
> then 2) turn to Scandanavian rally guys to use a
> mere wisp of that capability to give them their
> rally notes.
>
> Robin

Unknown Cabinet Minister "Thank God the Americans are finally in the War; they can always be trusted to do the right thing!"

"After they've tried everything else" W.S. Churchill 8 December 1941
>
>
>

---

John Vanlandingham
Sleezattle, WA, USA

Vive le Prole-le-ralliat

www.rallyrace.net/jvab
CALL +1 206 431-9696
Remember! Pacific Standard Time
is 3 hours behind Eastern Standard Time.

john vanlandingham Wrote:
-------------------------------------------------------
>
> Unknown Cabinet Minister "Thank God the Americans
> are finally in the War; they can always be trusted
> to do the right thing!"
>
> "After they've tried everything else" W.S.
> Churchill 8 December 1941

"And then the Americans rang up and asked if they could join in"

Anthony Hancock.. 'ancock's 'alf 'our

Ribbon

Robin: these accelerometers aren't gyro-based. They don't precess and drift. Where you think you are based on what they tell you might drift, especially because you'd have to smooth the data and would this create an error factor, but that's not a function of the accel, that's a function of how you interpret the data. Personally, I think it's nuts to try to recreate the road profile from an in-car accelerometer and odometer. Much better to use an external source such as GPS, and only use the in-car stuff as correction hints, or for the case where you can't get a good 3-sat GPS fix. And yeah, if I were doing this I'd convert the road to a vector of splines or something. I'd probably also do vidcap and audio cap so you could replay the road and your audio comments later.

DOS? Good god. Accessing serial/usb ports varies somewhat on different platforms, but the received data and what you do with it is the same. There're a bunch of ways to develop cross-platform GUIs, so you could theoretically write this so it could run on Win, MacOS, linux, FreeBSD, whatever. You could even use Java, I guess (Java annoys me). DOS is ridiculous.

---

Self-righteous douche canoe

hoche Wrote:
-------------------------------------------------------
> Robin: these accelerometers aren't gyro-based.
> They don't precess and drift.

I see your point (the system doesn't require knowledge of position or angular velocity - although you would think it would *like* to know attitude at given points). Gyro <> accelerometer - an inertial measurement unit is both gyro and accelerometer - based. Accelerometers are also subject to bias - the point I was making is that the quality of the instruments varies greatly, so to say that you have a 'state-of-the-art Wii' - (whatever state-of-the-art might mean) and can integrate into a measuring wheel and produce rally notes is a stretch.

(snip)

> to use an external source such as GPS, and only
> use the in-car stuff as correction hints, or for
> the case where you can't get a good 3-sat GPS fix.

Doesn't give you profile or very accurate horizontal - I mean, you could just do it from online mapping if you want, but it's the verticality that poses the problem

You're going to need either:
1) a minimum of 4 sats with good geometry all the time (we're doing this in 3D so 4 sats) plus it'll be differential in order to get proper vertical or, (more likely)...
2) GPS coupled with an IMU run through a Kalman filter (Novatel Span, or Applanix, or Crossbow) so that you can still get the profile under tree cover - still differential, but not always requiring full satellite coverage.

> And yeah, if I were doing this I'd convert the
> road to a vector of splines or something. I'd
> probably also do vidcap and audio cap so you could
> replay the road and your audio comments later.

It might be one of these situations where somebody invented something to solve a problem that doesn't exist.


As far as the cost goes - I think I understand why it costs what it does.

It's a choice between
a) high capital-cost black box + low labour cost
vs.
b) low capital cost Rube-Goldberg-have-to-manually-convert-text-files-to-three-different-formats box + high labour (whoops, I mean intellectual capital, right Dave ;-)) cost. (that was not an attack - imagine Julie Andrews in the Sound of Music as you read that)

I think the misunderstanding is that people wonder why it isn't low capital cost black box + low labour cost - but it seldom works that way IMO

I'm going to shutup now.

Robin

Robin Fleguel Wrote:
-------------------------------------------------------
> 2) GPS coupled with an IMU run through a Kalman
> filter (Novatel Span, or Applanix, or Crossbow) so
> that you can still get the profile under tree
> cover - still differential, but not always
> requiring full satellite coverage.

I wasn't thinking of using a real IMU. They're sort of spendy.
Xsens has a neat little guy that's a combo GPS and gyro, and
has onboard Kalman filtering, but last I saw it was about US$2K.

> It might be one of these situations where somebody
> invented something to solve a problem that doesn't
> exist.

Well, I can't argue with that. Might be kind of a fun project though.

> I'm going to shutup now.

Me too.

---

Self-righteous douche canoe

When I was preparing notes there was no GPS input to Jemba and I don't see how GPS would benefit that system. After measuring a stage, we'd view it mapped out on the laptop screen and it draw a 2 dimensional map of the stage. We'd compare to topo map or organizer supplied map to check we followed the correct route. Cool its draw such an accurate map by just accellerometer and odo.

When I installing digital TV and telemetry in NASCAR racecars and race school cars, had all kinds of issues with losing GPS signal, giving slightly incorrect position and the GPS speed given wouldn't be accurate enough to sycronize with accellerometer. I'd think any slight change in speed would affect the accellerometer readings. Remember the system has to dedect small crest while creeping down a road and distinguish between a proper crest or dip, bump, pothole, rut, ... Don't think about using GPS at all.

Can't say it can't be done but who is going to invest the time and $.

This story hasn't been told here so here you go. Canadian Targa competitor wreaks his vintage car because car ahead on road had crashed and had not displayed triangles. He spends a butt-load of $ and time developing the RallyCat

http://www.rallycat.ca/

thinking he could lease it for hundreds of dollars per team per event and thousands to organizers. Don't think he has made a dime and energy could have been spent more productively.

Dave Wrote:
> Canadian Targa competitor... developing the RallyCat

Cool system. I know that they used it at Targa at least one year. Haven't heard anything lately.

Anders

---

Grassroots rally. It's what I think about.

CommanderSalamander Wrote:

Hey Dave. Why would anybody name themselves online after a store in Georgetown? Is it still there?

Wilson

---

"Talk about drugs. Driving a car like that, going that fast, it’s like all the drugs at once." - Tommy Byrne

"Now, Pinky, if by any chance you are captured during this mission, remember you are Gunther Heindriksen from Appenzell. You moved to Grindelwald to drive the cog train to Murren. Can you repeat that?" - The Brain

wvonkessler Wrote:

> Hey Dave. Why would anybody name themselves online
> after a store in Georgetown? Is it still there?

How do you know they didn't name it after me? Don't know if its still there. Funny you knew about it though. Do you remember ElectroMax?

Thanks for sharing that Grant! I'd never seen any track building program, always thought about how cool that concept would be.

I was playing Colin McRae DIRT for Playstation 3 while I was home for the holidays, and unlocked the Pikes Peak Hillclimb. It was too real. I have a feeling thats what Monster used to practice for pikes peak, if not that than something very similar.

I'm assuming Bob's track building program is only for PC's right?

---

Michael LeCompte

Yup. Only for PC. The CMR Dirt PPIHC is close, but so far off at the same time. We're working on one though for Bob's now.

---

Grant Hughes

GPS is often used in inertial navigation to correct for the inherent drift you get when doing dead reckoning with accelerometers/odos etc. That being said, I don't know whether or not its needed for a jemba-like system. Given how cheap GPS is these days seems like a no brainer to use it in a localization system, even if you can expect dropouts.

How exactly does Jemba work?

The dork in me has been thinking about how I would make such a system since I read this thread. Seems like the obvious approach would be to break the task down to two steps. First, the system needs to be able to map the road. The simplest map would just be lat and lon, a step up would include altitude, a step from there would include crown and other angular descriptions. What level of complexity and the degree of accuracy is something someone with experience making notes could help answer. Funny thing though often people do things and can't really explain what they are doing in objective terms.."they just know what to do.."..and computers only work in objective terms .

Now that you have a trajectory of the road, that is a description of the lat, lon, altitude, crown, etc and depending on the complexity you decided upon, which I guess would be parameterized by distance from the start, you need some algorithm that can map those into notes. Of the top of my mind I could see geometric quantities such as radius of curvature playing a big role in turning all that info you collected in the first step into quantities that an algorithm could work with.

So, how does Jemba work anyway?





Edited 2 time(s). Last edit at 01/23/2008 10:00PM by Eddie Fiorelli.

Jemba is an ancient Norse magic.

Also:
1. jemba 5 up, 5 down

1. Word used to describe something extremely cool or amazing.

2. Used to describe a person that is extremely cool, awesome amazing, gangsat, sexy, smart, funny.

3. Also used to describe a person that has done something amazing or cool.

"man u shud hav seen this shit its was jemba"
"That was so jemba bro"
"man ur like a junior jemba"
"Dude ur so jemba rite now"
"man u macked those chiks like jemba"
"damn ur a genius man just like jemba"
"shit bro u dunked that like jemba"
"hey man keeping it real like jemba?"

---

RichardM



Edited 1 time(s). Last edit at 01/24/2008 11:53AM by Richard Miller.

I went digging around and wanted to revive this thread, because I thought our experience would be interesting to you all.

We wrote our own notes for NNR, and from the competitor feedback we got, apparently they turned out pretty well. What's more, we OWN the notes, which
means we can use them in future years, tweaking them as appropriate.

What's needed:
a) An index of turn degree to circle radii
b) Large open area to calibrate the test vehicle
c) Rally Odometer
d) Computer with Excel or other spreadsheet software
e) Lots and lots of time

Steps:
1) Drive around in circles in the large open area. First, mark off the distance of a circle indicating your "1". Drive around, noting the steering angle. Put tape on the steering wheel. Now do your "2"s. Then your "3-"s, "3"s, "3+"s. Put tape on the steering wheel marking each range. For bonus points, keep track of the angles on the steering wheel so you can reproduce this in that vehicle later.
2) Drive the course very slowly, down the middle. Have at least two people in the car. The driver is calling out the steering angles and the stop/start points of each turn. The passenger is recording everything, and making note of the odometer readings. The team argues until it reaches a consensus on things like whether it's one long turn that widens and tightens or two separate turns. Expect lots of argument. You will probably also have to stop and back up several times, so make sure you have TSD experience... We found that the average speed was pretty consistently about 7mph, so to do 40 miles of stages it took roughly 6 hours.
3) If you have three people in the car or lots of experience doing this, you can put in extra notes like "offcamber", "exposure left", and do things like link turns. Otherwise you will probably have to drive the stage again to get such details. I recommend doing this as a second pass at about 20mph. For 40 stage miles this took about another 2 hours.
4) Go home (or back to your hotel) and process the notes, putting them into a driveable format. This will take several hours.
5) Go back out and test the notes with a codriver. Fix any errors, link turns with underlines, add in extra commentary, and so on. In our case, we ran each stage twice, at an average speed of about 50mph. This was about 75% of rally speed.
6) Go back and make more corrections, and print up the final copy.

Problems we found:
1) The team HAS to agree on a vocabulary, and make that clear to the competitors. What's a jump? What's just a crest? What's a small crest? What the difference between a dip and a wash? How much exposure is required to put in "exposure"? What makes a "no cut"?
2) For the faster turns, the "end" of the turn at higher speeds was not the "end" of the turn at lower ones. In particular, the runnout at the end of a turn resulted in us having to shorten some of the distances between turns after testing.
3) It takes a LOT of time. If you count all the hours spent generating notes for our 40 miles of stages, it came out to something like 12 hours for two people on stage plus and additional 8 hours for one person to prep and format them. So, it was something like 32 man-hours total.

If you were to do this as a business, you'd have to throw in gas, hotel fees, and hourly wages:
Gas: 4 runs at 40 miles @ $3.50/ga: $560.00
Hotel bills: 2 people, 2 nights: $140.00 (ballpark)
Hourly wages: 32 man-hours @ $50/hr: $1600 (and that's probably cheap)

That's $2300, and doesn't include printing/binding costs, airfare or car rental, or gas to and from the stages. Nor does it include overhead for business expenses.


So, in summary, I now have more sympathy for JEMBA and P-Sport's prices. For a National, it's probably worth it to pay a premium to have a company come out and write notes, largely because their techniques are standardized. However, I do believe that regional organizers can replicate similar notes at a smaller cost. It just takes a lot of time. However, regional organizers can offset the initial effort by reusing notes in future years.

-michel

---

Self-righteous douche canoe

We write our own notes at Canadian events. Most people do. No computer needed. And, it's much quicker.

Use jemba or another computer "solution" if you feel the need to spread blame.

---

Ted Mendham
www.rensport.net