Measuring Slip Angle using accelerometers

[Twonius]

People who went to Claude Rouelle's seminar last year might be able to answer this one.

I went back in 04 and remember him talking about measuring front/rear slip angles by placing accelerometers at the front and rear of the vehicle. Does anyone know if this actually does/doesn't work? I asked one of our guys who went last year and he said that CR said it doesn't work (I don't remember him saying it doesnt work, only that it was possible). Anyone have more info on his reasons for this?

[Twonius]

People who went to Claude Rouelle's seminar last year might be able to answer this one.

I went back in 04 and remember him talking about measuring front/rear slip angles by placing accelerometers at the front and rear of the vehicle. Does anyone know if this actually does/doesn't work? I asked one of our guys who went last year and he said that CR said it doesn't work (I don't remember him saying it doesnt work, only that it was possible). Anyone have more info on his reasons for this?

[Jersey Tom]

You could figure out O/S and U/S by doing that..

Slip angles though.. I don't know. I'm not a vehicle dynamics guy. But depending on your suspension/steering/ackermann setup your tires are all going to be driven to different slip angles. Can't imagine how youd extrapolate slip angle at all 4 tires from just two lateral acceleration measurements.

Might be able to get an average accross the axles, but I think you'd need steering angle input and tire data correlating lateral force to slip angle.

Just man up and buy the $20,000 slip angle sensors. Gosh.

[^_^]

Are you just trying to get your post count as high as you can? You and HomemadeWRX both.

[GSpeedR]

Integrate the accel output at each axle and you have a lateral velocity. This divided by the longitudinal speed of the tire is your average angle for the axle. If you want to break it down between tires on an axle, you'll need to plot you toe change from bumpsteer/rollsteer and your steering pot (w/ackerman) at the front. Next, find a pen or two and start signing autographs for all the people who want to be your friend.

[Twonius]

The subject actually came up in my vehicle dynamics class. Our prof had done some research on the subject and they'd had trouble getting the results to converge (integrate vehicle rotational velocity to get a full roation over a lap). I'd vaguely remembered something from CR but I couldn't peg wheather he said if it was possible or impossible.

the reason given in VD class was that the calcualtions become unstable when noise is introduced into the system, and it wasn't a differentiation.

[GSpeedR]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Twonius:
The subject actually came up in my vehicle dynamics class. Our prof had done some research on the subject and they'd had trouble getting the results to converge (integrate vehicle rotational velocity to get a full roation over a lap). I'd vaguely remembered something from CR but I couldn't peg wheather he said if it was possible or impossible.

the reason given in VD class was that the calcualtions become unstable when noise is introduced into the system, and it wasn't a differentiation. </div></BLOCKQUOTE>

I'd rather use yaw rate anyway. The noise from accel would probably cause a significant bias that you would have to consider/filter. I don't know if that's what he means.

VD class? That's a class you swear you've dropped but suddenly find yourself attending regularly.

[John_Burford]

It comes down to the yaw sensor not being sensitive enough to charaterize the slip angle behavoir.

John Burford

[Andrew Nabb]

I am not sure commercial GPS will be accurate enough to use as a slip angle measurement device. At Work (toyota racing development) we have one optical slip angle sensor, which cost huge bucks. It uses some kind of machine vision to "watch" the grain in the asphalt as it moves by.
Maybe if you had acess to the goverment style high accuracy gps you could do it.....however since the optical sensor we have costs several thousand dollars (60 grand i think) i would be tempted to think it isn't that simple.

[Alan]

I would try bolting an optical mouse on each wheel.

[Buckingham]

Mr Nabb,

You could try a VBOX at each end of the car. That would only cost you $14k each. http://fsae.com/groupee_common/emoticons/icon_smile.gif

[murpia]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by DohertyWins!:
Integrate the accel output at each axle and you have a lateral velocity. This divided by the longitudinal speed of the tire is your average angle for the axle. </div></BLOCKQUOTE>

Sorry, not sure this works. Slip angle generates the lateral force on a tyre but lateral force does not imply a lateral velocity. Imagine a cross-wind situation, the tyres will be developing a force and therefore be running at a slip angle, but no lateral velocity is involved.

Ian

[js10coastr]

mathematically and theoretically, by using accelerometers at the axles you should be able to get slip angles.

The problem arises when the noise and vibration in the car and DAS is logged and then integrated. The noise/error is amplified in the integration.

Solution: Call OptimumG and use their rental loom setup with 2 slip angle sensors and tire temperature sensors. While you're at it you should probably sign up for their seminar too...


This message has been brought to you by OptimumG... it's Edu-tainment.

hehehehe

[js10coastr]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by murpia:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by DohertyWins!:
Integrate the accel output at each axle and you have a lateral velocity. This divided by the longitudinal speed of the tire is your average angle for the axle. </div></BLOCKQUOTE>

Sorry, not sure this works. Slip angle generates the lateral force on a tyre but lateral force does not imply a lateral velocity. Imagine a cross-wind situation, the tyres will be developing a force and therefore be running at a slip angle, but no lateral velocity is involved.

Ian </div></BLOCKQUOTE>

...but when the wind hits the car, there is a slight delay until the car can respond, and the tires generate a slip angle. During this time, the accleration changes. The acceleration signal is integrated, and that change in acceleration is seen in the generated velocity channel. It is hard to keep in mind, but what we are measuring is the acceleration, and not the velocity.

Hope that make sense...

[Test Driver]

Hello,

I tried working total vehicle slip angle from one Vbox before. The accuracy is not there with the instantaneous radius bouncing all over the place.

Alan's idea sounds very interesting to me. It does seem that optical mouse devices work very much like Correvit sensors do.

Ramon

[mtg]

The problem with optimal mouse sensors is that they only work when they are within a couple millimeters of the moving surface. How stiff is your suspension (and tires)?

Also, if you look at the specs on the Correvit sensors, a lot of the price probably is in getting the accuracy of those sensors- they are damn accurate.

Datron just came out with new Correvit sensors that are about half the size and weight of the old ones. Price? Probably not half as much...

[Alan]

My comment about the optical mouse sensors was a joke. My understanding is that they use similar technology. But I doubt you could modify a mouse to actually to do anything useful. But then again I have never given it serious thought either.

[Twonius]

Okey heres something I just thought up that probably wont work but..

If you have a good tire model, couldn't you instrument your suspension links, solve for the lateral force and estimate that tire's Fz based on lateral acceleration, roll angle, roll distribution etc. Then once you have Fz and Fy, backsolve for slip angle..

This sounds very problematic based on the fact that tire tests are done in labs.. at controlled temperatures and surfaces so you'd probably just get garbage, but maybe it'll give someone else a good idea.

[ben]

The tyre model wouldn't be that accurate. The better way round would be to measure the slip angle properlay and back solve to create the tyre model. This would allow you to account for surface effects. If you can't measure the slip angle I wouldn't bother going to that level of effort.

Onboard video of the car cornering on a grid marked road might help - read the Milliken publication of Olley's work.

Ben

[Test Driver]

I would pursue the mouse idea. There's lots of technology out there that was the sole domain of race teams and big OEM/Suppliers that should be accesible to the masses.

This sounds like it could be a good project for one of sparkies on your team.

Ramon

[ben]

I believe Monash tried the optical mouse thing and I don't think the resolution/accuracy was good enough.

Ben

[Jarrod]

yes, Nick at monash tried to amke slip angle sensors from an optical mouse, even went to the trouble of making a disc of "road" that he put in a lathe, then moved the mouse to different distances and used different speeds, to se the outputs. If you play around with the lens you can get it to work at longer distances, but it was limited to about 15-20km/h if i remember right, and the resoolution dropped off at higher distances. With a better mouse and some development on lenses it may still be possible. We also looked at small microwave doppler sensors, about 20 dollars each, and I think further research down that path could come up with something useful.

[Twonius]

The mouse thing got me thinking, Why does the sensor have to be non-contact?

why cant you use a caster (spring mounted)with an encoder on it to measure slip angle. The caster wheel will naturally want to move in the same direction as the ground plane and this difference can be recoreded using the encoder.

Now I'm definately not an expert in high speed caster wheel operation, but i do remember that they tend to spin at high speeds as they skip across the surface and loose lateral stability, What would be the solution to this? Rubber tired caster, bigger caster, or spring/damper mounting it?

Once again its a simple idea but casters are cheap and it might be worth a shot.