Update - Bicycle Anti-lock brakes Part 2

I think you need to study the very early 1st generation ABS systems ever put on production cars and model the system after that. I believe the systems employed today are highly coupled to the stability control systems which attempt to use individual wheel braking and sometimes throttle to control the yaw of the vehicle as well as prevent sliding. The oldest systems were far simpler and probably more documented. I'm sure a lot of what they do now is proprietary and trade secrets. It's fun to watch your experiments. Just don't go over the handlebars!

U.S. Water Rockets

From the research I've done, it's very difficult to say how the car ABS actually works. I was just making an assumption as to which system is similar to my testing. Here is a research paper if you wish to read up on the many systems: <a href="https://curve.coventry.ac.uk/open/file/aebcb301-fec5-4684-945c-020817157509/1/Chen%20Li%20PhD%20Thesis%20_Redacted.pdf" rel="nofollow noopener" target="_blank">https://curve.coventry.ac.uk/open/file/aebcb301-fec5-4684-945c-020817157509/1/Chen%20Li%20PhD%20Thesis%20_Redacted.pdf</a> I was making the assumption that the car in my testing was using something similar to the bosch system, which you can read about on page 63. The question "How does ABS work?" certainly seems to be a 'how long in a piece of string' question.

Tom Stanton

From my understanding, (not heaily educated in this stuff) a car uses all 4 wheels as refrence, finds avg deceleration and then compares each wheel vs the avg, if 1 wheel dips too low (or if all dip to fast) the system lets off pressure. As the goal is not only to prevent slip, but also minimize chances of spin out when stopping fast. (When in conjunction with TCS)

Giovanni Viscardi

Yes the accelerometers are noisy for sure! I've been experimenting with the arduino kalman filter library and it seems to work very well. I've done a few tests today and the extrapolated velocity isn't far from the actual velocity, though it's only occurring over a couple of seconds and always experiencing a hard deceleration. As soon as the bike stops, it drifts away. Thanks Gavin!

Tom Stanton

I'm not sure cars behave as you describe. Perhaps it depends on the car manufacturers doing things differently. I don't believe they look at one wheel to sense the speed of the car simply because each wheel could be on a surface with different friction. Say one wheel hits ice or 3 wheels hit ice. I would think that the system applies each wheel independently (some cars have only 3 modulation channels, where the rear wheels work in unison). The system should only be aware of the wheel locking up and then once it has a reference point it can modulate the pressure up and down and see if the wheel locks, in which case it backs off until it begins rotating, then increases the pressure until it locks and then backs off quickly. You don't perceive the wheels as locking up on video because it stops and starts so rapidly.

U.S. Water Rockets

Makes a lot of sense, exciting! :)

I'm using a regular open loop stepper, but in hindsight I probably should have got a closed loop stepper, even a servo motor. Yes I agree with you on that, I would be very surprised of someone could lock all 4 wheels on dry/wet tarmac without any lubricants like oil, ice or leaves. Which is probably why a car has atleast one reference wheel to measure the speed of the car. Then I assume in a worst case scenario, if all 4 wheels were to lock, it would measure a large deceleration on all wheels and release all the brakes. Thanks!

Tom Stanton

I love your commitment to getting this working. I am honestly interested to see how this turns out. I was actually looking into extrapolating velocity from acceleration the other day using an arduino, but my concern became one of filtering. (Those little accelerometers can be very noisy, as I'm sure you're aware of.) Just something you might want to watch out for. I know there are some digital signal processing libraries available for arduino if you need to do filtering. This might be old news to you, but just thought I'd give my 2 cents :)

Gavin Remme

Yes its far more complex than I initially thought when planning the project! Thanks

Tom Stanton

Hey Andre, yes that makes complete sense! I didn't explain my concept very well, but it's not far from what you explained. The wheel deceleration 'activation' limit is only for the initial locking of the wheel to start the ABS loop, similar to how the car initially skids and then the ABS kicks in. Then once it's confirmed the wheel has lost grip once, it'll use this as the reference point to monitor the slip ratio. In the video I want to test three concepts just to show what works and what doesn't. The first will release the brake when the wheel speed is zero, which is obviously too late. The next will be wheel deceleration limited (like you explained initially), but I expect this can only be setup to work on a particular surface as deceleration on grass is far from deceleration on a road. Then the third is the wheel deceleration triggered slip ratio control loop. Thanks!

Tom Stanton

Hey Tom, thanks for the update, very interesting as always. I don't know if it helps, just troubleshooting in my head, but on your proposed solution to implement going forward to use accelerometers on the bike frame to measure when the deceleration is bigger than a set value activate ABS. Sounds like it should work for testing but maybe not for actual useful ABS as it might be a self limiting system? Meaning the purpose of ABS is to slow down a vehicle as fast as possible with the available grip? so if it triggers early before the wheels skid you will be losing stopping performance and conversely if you brake so much that the wheels skid then at that point you start to decelerate at a slower rate and hence ABS wouldn't be triggered either? Not sure if this is right at all but in my head where I would go next with it would be to monitor deceleration on the frame as you propose but compare the slope of that graph through time to the brake lever position. Meaning if you are braking a little (20%) and your deceleration is small (eg -1m/s) all is good. Then you brake more (50%) and you decelerate more (-2m/s). Then you brake harder still (90%) and a wheel locks and because of that you don't decelerate as fast (-1.8m/s) so then the ABS system kicks in (as the slope of the graph has changed signal) as there a big gap between what your hand wants to do on the brake lever and what the bike frame is actually doing (deceleration) and then the job of the ABS system modulating the brakes is to increase the frame deceleration as much as possible. Again don't know if it makes sense or would work but just my 2 cents. Keep going this is awesome! (edit: it's late, sorry for any typos, long sentences, etc)

Wow Tom, you have been very busy. Thanks for explaining the car ABS. Never thought about how they actually work. Researching a project isn’t always as easy as just typing a question into Google. Keep at it- you’ll get there. Cheers!

Jeff LaHay

Tom I wondered if during your research you came upon the Lucas stop control system. It was a mechanical system controlling the hydraulic brake pressure and was fitted to some front wheel drive Ford cars in the nineties. Might give you inspiration. <a href="https://www.howacarworks.com/illustrations/lucas-girling-scs-stop-control-system" rel="nofollow noopener" target="_blank">https://www.howacarworks.com/illustrations/lucas-girling-scs-stop-control-system</a>

Bravo Tom for even attempting this project. Please continue with update videos like this to explain technical hurdles you are trying to overcome.

Very cool Tom! Thanks for the inside scoop.

Fishysan

research cost time, was not expecting a update.

Regarding ABS: Without emergency braking assistant it is actually not easy for an unused person to break fully. After several drive safety courses i learned that most people just brake sufficient to have one or two wheels braking 100% into ABS, but rarely all 4 (If one is into ABS people believe they reached the max because ABS starts to work, but actually it could only be one wheel, this is probably why emergency braking systems were developed detection the intention to fully brake and helping to really do that :-)). You have to push pretty hard to get all 4 wheels into full braking. Typically you would see the wheels lock or at least create more slip from time to time as the system tries to find the best level. And below 5 or so kmh ABS is off so you would see the wheel block just before standing. Not saying the braking video is wrong, just challenging whether the brake was hit really hard :)

How do you drive your stepper motor? Closed loop would give extra torque, especially at zero or low rpm.

Wow! It's complicated, isn't it? Good on you to get this far. Keep going.

Peter Rogers