Traction feedback device?

[Benk]

Hahahaha, just had a read of this thread. Fucking hilarious. Cheers for the laughs guys

[bogan]

I think you're underestimating just how awesome humans are in terms of sensing stuff. The human is an extremely complex system of accelerometers, gyros, force and pressure transducers, optical and audio sensors all interlinked with powerful and precise actuators that create complex feedback based on actions.

So, you want to estimate coefficient of friction of a tyre based on a brake test, you'll either need to know or assume:
Mass of bike
Mass of rider
Centre of gravity position of bike
Centre of gravity position of rider (and assume he doesn't move.....)
Friction coefficient load sensitivity of tyres (and assume a curve shape for it, because it's not even close to linear)
Ellipse ratio of tyres - that is the ratio of max lateral force to max longitudinal force (this and the load sensitivity will give you the most crude tyre model that would allow you to estimate available force within 5%)
Wheel rate of front and rear suspension (to allow you to estimate forces in the highly dynamic scenario of braking), this could be calculated if you have suspension displacement pots
Damping ratios, front and rear, along with the force/velocity curves (I don't know how you'd get these without a shock dyno)
The values of anti-dive/lift that the suspension has designed into it

So, with the above you could potentially work out a coefficient of friction correction factor for the front tyre for the surface the test is performed on. We still have no idea about the rear tyre so we'll have to guess that I suppose. Let's try and apply it to going around a corner

First we'll need to correct the coefficient of friction for temperature, if the rider did the brake test on colder tyres the grip levels will have changed. You could assume it's negligible or characterise a temp vs friction curve and measure temp with an IR tyre pyrometer (solidstateracing.com has them for around $300NZ)
Next, we need to know what NORMAL load is on the tyre. We're going to have to assume the rider hasn't changed his CoG since the brake test, then we can use an accelerometer to work out loads. This gives us an estimate of lateral/longtudinal force that the tyre can currently supply
Then, we need to know how much longitudinal force the tyres are generating. This is going to be hard - we can assume a split based on mass distribution for lateral, but not for longitudinal. We can use an accelerometer to work out the total longtidunal acceleration, but unless it's a positive value (accelerating rather than braking), we can't easily tell which wheel is causing the deceleration. This is where stuff gets tricky - we could look at brake pressures, but that wont help is the bike is engine braking, or getting some drive from the idle. Not sure how you'll solve this one, but it needs to be solved because you can't ignore the effects of combined braking/cornering on the force a tyre can produce.

Let's say you've got the above sorted - it's all very nice but we've still got some big assumptions to cover.
Road camber - We could ignore this, but I think that would be a bad idea because I'll bet off-camber corners are to blame for numerous crashes. We can get road camber using 2 optical proximity sensors on the bottom of the bike, spaced a known distance apart, provided we also know the lean angle of the bike relative to horizontal.
Surface irregularities - you've stated earlier that suspension is supposed to maintain constant force on the tyres, and therefore you can neglect this. Hmmm. When you hit a bump there is an increase in normal force on the tyre - this force accelerates the unsprung mass and compresses springs and pushes fluid around in the dampers. It also accelerates the sprung mass by some amount. The amount the force increases by is not particularly easy to work out, infact it's damn hard. The amount it decreases by on the other side of the bump, as the wheel is returning to normal position - that's the real problem, and that's where you're going to lose grip. The force variation is dependant on: sprung mass, unsprung mass, spring rates (and motion ratio), damping force/velocity curves, tyre spring rate, tyre damping, and any hysterisys in the system due to friction (and probably more stuff I don't know). To give you an idea of the magnitude of these forces, you'll probably see a peak force on a good bump that's around 2 to 3 times greater than the force riding over smooth surfaces

There's more assumptions that are going to introduce error into your system, but I'm not here to berate you.

Now, as a final note -have you been tallying up the costs of the sensors listed above, and the time to install and calibrate them correctly onto the bike?

I realise the human system is very good at sensing data, but the brain has to process the data to decide what was actually a safe speed or not, for a rider that has never been close to the limit of traction they will have no idea what those limits are, and could be grossly over/underestimating them, this system is designed to provide an approximation of those limits to the rider.

For the list of assumptions, it is simple to get all of them except for the suspension stuff, which i am leaving out of the model and assuming ideal road surface conditions for the initial design. This is because suspension action is closely related to the roads surface parameters, which we cannot know without expensive sensors, and i want to see what can be done without getting too carried away with expensive sensors. Im well aware it may be extremely limited without them, but it also may not.

The system is designed for road use so i think the temperature change could be ignored, or assumed ambient + a few degrees.

The longitudinal acceleration is one of the bigger possible grey areas, but for cornering the system should be able to judge between a slight engine braking and a brake being applied, and assign the force values to the tyres accordingly. (can also easily be wired to the front/rear brake light sensors too)

The road camber and surface irregularities technically are neglected by the system (as it is too difficult to reliably sense these). But i didnt say they could be neglected in calculating the actual traction available, the device calculates the maximum traction available in normal conditions (no camber, no bumps, no slippery spot or wetness) and the rider is responsible for seeing this and adjusting speed to compensate.
It is worth noting with respect to the surface irregularities that while you do get big force spikes etc, the suspension is designed to limit the effect these actually have on traction, its still there of course but not greatly reduced from the 2-3 times felt at the wheel.

One of the constraints i put on the system was to limit the cost and number of sensors needed on the bike, another reason why suspension and road surface is not sensed.

On another note im starting to think that an intermediate step would probably be a good idea, so i will probably try and create some modeling software and get a better idea of just how much all these factors (which i belive will not generate significant error) actually contribute to the system, just to make sure

[bogan]

Hahahaha, just had a read of this thread. Fucking hilarious. Cheers for the laughs guys

alway happy to increase the lols

[CookMySock]

The sensors and other hardware is dead cheap. Its the thousands of hours coding and doing the maths that will hurt. Maybe as you say - start off installing cheap accellerometers and high-resolution ADCs, and harrass the chassis a little and see what humps you can make appear in the graphs.

I think the road surface, suspension, and tyres can be ignored. What will be telling are the little sideways slide-recover-slide events.

edit: http://www.google.co.nz/search?q=dspic

Steve

[bogan]

The sensors and other hardware is dead cheap. Its the thousands of hours coding and doing the maths that will hurt. Maybe as you say - start off installing cheap accellerometers and high-resolution ADCs, and harrass the chassis a little and see what humps you can make appear in the graphs.
Steve

yeh the stuff ive got on the first one (essentially a lean angle indicator) cost about $30, and displays the cornering g-force fairly accurately

[alanzs]

You could build a device, so that it watched how you were cornering, and if you gave it a fright it could stand the bike up for you - thereby saving you the bother. That would be easy to build, and you could have it err on the side of "safety" (sic).

Steve

You could just put side cars on both sides of the bike? Would that be a trike?

[Malcolm]

ok well, while I'm still in strong disagreement with a lot of what you've said, it's obvious that you've made up your mind and wont be convinced otherwise. Good luck with the system - stubborness is something you'll definitely need if you're going to make it successful.

[bogan]

ok well, while I'm still in strong disagreement with a lot of what you've said, it's obvious that you've made up your mind and wont be convinced otherwise. Good luck with the system - stubborness is something you'll definitely need if you're going to make it successful.

Im not trying to be pigheaded about this, but the fact is there are very few facts on this subject out there, so all ive got to go on is personal opinion, of which i favour my own of course!

If you guys are right, itll still be interesting to see (and quantify) just what effect all the different factors actually have on traction.

[Malcolm]

haha, thanks for the negative rep points whoever that was

[bogan]

not me, i never heard of them till now, they seem like a good idea though

[Hoon]

ok well, while I'm still in strong disagreement with a lot of what you've said, it's obvious that you've made up your mind and wont be convinced otherwise. Good luck with the system - stubborness is something you'll definitely need if you're going to make it successful.

I think the problem is that you both have different expectations of the final product. Malcolm, yours is the ideal dream solution that we'd all want if we could, no expense spared, no stone left unturned, with the highest possible accuracy based on the technology and knowledge available today.

Bogans is the simplified, low cost, student kiwi no.8 wire shed version obviously with a number of compromises, limitations and of course sacrifices in accuracy. Sure it's far from ideal and the first n versions will probably be arse but you've got to start somewhere. With the right encouragement and support it may one day evolve into that ideal tractionomometer we're all after. Who cares if it doesn't when all it costs me to find out is a few encouraging posts.

[motorbyclist]

TL;DR

I reckon that as it can't anticipate things like oil spills and gravel etc and other variables, combined with idiots who will beleive it and forget about those variables and those who will look down rather than forward, this product would most likely cause more injury than it prevents.

of course a cheap telemetry system would be a cool gadget, but they're nothing new either


and who is going to test eacha nd every tyre out on the market and account for variations in rubber quality and road surface?

[CookMySock]

I reckon that as it can't anticipate things like oil spills and gravel etc and other variables, combined with idiots who will beleive it and forget about those variables and those who will look down rather than forward, this product would most likely cause more injury than it prevents.

Pray do tell? What thing, human or machine, could ever do this? Look around corners for gravel patches and other slippery bits? I do not think so. On NZ rural roads, the bottom line is riders are going to be outriding your corners. So if you round some corner and spot a said slippery bit, nothing except your wits and will to survive will save you, and certainly not some box of tricks with lights on.

of course a cheap telemetry system would be a cool gadget, but they're nothing new either

Yeah this thing would do that.

and who is going to test eacha nd every tyre out on the market and account for variations in rubber quality and road surface?

You wouldn't, and couldn't. The unit is really only interested in bad things that happen to the bike overall - that is a fairly predictable science - or at least there is some point in analysing it.

TL;DR

No thank you, I'm not that sort of boy.


Steve

[cowpoos]

haha, thanks for the negative rep points whoever that was

Geee...I wonder who that could be?? Steve? you got a big enough backbone to own up??

[Malcolm]

Ha, I'm not too bothered about it, just thought it was a tad petty because I wasn't being nasty or anything - but you can't expect much better from people on the internets

My concern is that the system is far too complex to simplify and still get good results out of - but I guess that's something I can only speculate on and you'd really need to build and test something before you'd really know how useful or accurate it was. And of course you'll also need to decide how accurate is considered acceptable. I think Dangerous bastard is on the right track with looking for the telltale signs that the bike is about to lose it (presumably the same way a human works when they're good at riding at the limit), but that's not going to give you an incremental measure of how close to being fucked you are, it'll just suddenly say "oh shit, that was close"