This was fascinating to watch. The biggest improvement to motorbikes over the last 20 years has been in brakes, and I've often wondered how difficult it would be to create a retrofit kit to fit to older bikes that are still more than fast enough, but stop like a grayhound on a polished floor. One of those things that seems conceptually simple, but evidently complex in practice.
Why even in commercial devices the brake is released so significantly? The best choice would be a PID control to release slightly the brake to get the best possible deceleration according to tire friction, the deceleration would appear as constant (the PID loopback can be done thousand times a second).
It is not released that much in commercial systems. In a car you have three states build up pressure, hold pressure and release pressure. The abs changes between these cycles 15-20 times per second. In most situations this is more than enough.
@@NathanaelKeller My daily driver has one of the oldest antilock systems Ford put in a commercial vehicle, and it's very slow at just above 4 cycles per second. From experience, I can tell you that it's plenty good enough to keep traction even in heavy snows, even if it's perhaps just a bit too aggressive, increasing stop distance just a bit. It does sound like an old, heavy machine gun though, so that's cool.
Absolutely the first thing which came to mind, PID or PD. Best way to decrease overshoot and minimize rise time since the wheel speed and acceleration is changing dynamically with the surface contact. Completely unnecessary on a bicycle, but an enjoyable design process.
@@Nolan001001 the trouble with a PID for this specific situation is it needs to be highly tuned which is possible if you have consistent parameters, but an ABS system needs to work across a wide range of grip levels, so it can't be tuned easily. In reality, the software behind real ABS systems is highly sophisticated taking in multiple sensor inputs and determining what is the maximum grip that can be achieved. A lot of modern systems also now include active stability control which will brake the vehicle unevenly left-right and front-back to try to help prevent spins.
@@lloydmorrissey Very true, I was mostly speaking related to the project in the video itself, how to get this specific application to work "better", like I said previously, would be fun to do. And besides I'm not a control systems engineer, I wouldn't even begin to fathom the block diagram of a true abs system and the resulting transfer functions.
Physics is the greatest teacher of brake control on a mountain bike. Crash enough times and you'll figure out what you're doing wrong to fix it. Ideally, you'll be a one trial learner.
Skid the front wheel for 0.2 seconds you'll be on the deck, too - for the system to work it would have to somehow prevent any skid from even starting...
I was going at 35km/h. Suddenly a person came in front. I have rim brakes and very bad tyres. And also don't have a front brake. There was little space from the side so couldn't go in speed. Did full emergency brakes tyres skid and turned to the left, then right and then left. What a drift. Was so thrilling but wasn't intentional.
Yeah I just now threw in a comment about how he needs a huge motor because he's trying to counter-act his hand, and he's trying to do it at the handle instead of the brakes. With the right mechanism, you could use a solenoid instead of a servo, and have response times in the 10s of milliseconds.
Rather than having an ABS system that releases when it senses a locked wheel, would it work to make them completely fly-by-wire where the brake lever is just an input to the computer to indicate percentage of braking force? At 100%, the computer would only apply the maximum force that would still keep the braking of the wheel within the ideal slip ratio portion of the graph. So, instead of trying to modulate the pressure after it's locked, only enough braking force would be applied to maximize deceleration without the lock up.
This is just fantastic. This video had me thinking about all sorts of things including why the ABS judders your wheels in an emergency braking situation.
The juddering is mainly noticible in older systems that could only cycle between full clamp and full release. Repeat this nearly 100 times per second and you induce some fun vibrations.
Your data should also allow you to calculate the friction of the surface from the first lock. You can then use this to predict the maximum safe breaking force and release the break to the ideal position. You'd just need to adjust the break pressure curve to allow the wheel to speed up again before clamping to the ideal pressure.
Alec Parkin I was greasing my hub, didn’t realise I got some on my disk, tested me hub and this used by breaks to slow only to realise I have got grease in my pads
Well done mate! It really shows how hard is to replicate such easy to understand mechanisms. The first brake release mechanisms looked way cooler though ;-)
@@samuelyoung2671 or a smaller stepper, and hydrolic brakes. ... and the 120slits cut into the brake rotor itself. . and an AI camera system too create a speed differential, - and of course, do it on both wheels. ;-D. totally fantastic tom !
@Lassi Kinnunen its really quite exciting, for a push bike, - and i thought an electric motor was cool.... wtf. - engineering pwns * checkout my "free energy motor* demonstration ruclips.net/video/sJ3zjje8hgw/видео.html
Very cool work! Perhaps a solenoid or other linear actuator would have better mechanical response time than the motor. Very good points about real-world applications too
I think the reason automotive ABS uses hydraulics is the immediacy provided by incompressible fluids. The exact moment the actuator is initiated the effect is applied. Practically zero latency.
Long before ABS, as a professional emergency services driver (fire) we were taught cadence braking; essentially manually operated ABS. I used that when driving my own car pre-abs and do the same with the front brake on both motorcycle and bicycle. It is an acquired skill but still very effective. It works best on hydraulic cycle brakes but I have also found that the front disc is so easily contaminated with road muck that it effectively turns itself into ABS by alternately slipping and locking but it is very noisy! I used to clean my front disc and pads very regularly but found it locked up on gravel far too easily so I left it alone. It works!!!! It's noisy but very effective (self) cadence braking. No need for expensive ABS as long as you can stand the noise plus the noise itself turns you into a very careful breaker; effectively a verbal cadence braking warning. It works for me and has done for many years. I'm 71 years of age and still riding 100+ miles per week safely. 😀
@@Gunzee honestly anything that wasn't a servo motor would have been 100x better. There was no need here for precise control (he connected the thing with a piece of rope)
Glad you ditched the first prototype, as you were running into the prpblem with this being a cable based abs braking system. When you look at a cars abs system, it uses hydraulic brakes and a very expensive abs pump to quickly and precisely pump individual brake circuits. You want the fastest possible mechanism controlling the pulse and release part, like a solenoid or high end RC servo. These slow stepper motors arent cutting
The Bosch E Bike ABS also uses a hydraulic valve to engage and disengage the brake, they just don't use a pump feeding back the hydraulic fluid, you have to do this manually with the lever.
That's one of the videos that made me retrofit an ABS on my bike (which never had it as an option from factory). Oh and later upgraded it into a traction control system with 4 different modes. And it works like charm - saved my ass a couple of times. Thanks for an inspiration!
Accelerometers are useless, theres no way to know between a locked wheel and you breaking slowly. The only way would be if you correlated the accelerometer to the wheel decceleration, and activated the ABS if they get too far away.
Older ABS systems did not use accelerometers. The really early systems were entirely mechanical, have at look for the "Lucas Girling Stop Control System", a very clever mechanism.
Are you sure the motorbike uses an optical sensor for the wheel speed? I've only ever seen vehicles using hall effect sensors, since the optical sensors would be subject to dirt and grime blocking the light emitter or detector. I've been looking forward to the next video, so glad it came out! Another fine video presentation. Nicely done!
Yeah, the MC is hall effect sensor (edit: Induction sensor actually). Never seen one done with optical. Car or MC. Or anti skid on aircraft for that matter. That's where it was original developed btw
I wasn't 100% sure, but just assumed as it used a slotted disk. I would have thought a hall effect sensor would have a solid disk with the magnets embedded inside?
@@TomStantonEngineering It is not even hall effect sensor. The slotted wheel is called a reluctor ring and it induces a current in the sensing coil (in the sensor).
I really appreciate your analysis of *whether* and *for whom* ABS might actually make sense on bicycles. Requirements engineering is a key part of engineering!
i was thinking along the same lines thinking linear actuator opening a chamber like a piston but it would have to be small enough strong enough and quick enough
@@oldskoolhead0 Yeah holding force might be a problem ,but I think a piston on a linear actuator would be quite hard to move. Especially if you kept the cross section down to limit the force that could be applied...
@@swamisamantharella i suppose it could be put on a cantilever like the brake lever itself or maybe even put on the end of the brake lever between the handlebars (which would also mean its normally closed instead of normally open, so no need to hold anything for too long either) and it would only have to fight against your hand
Hi Tom. I see that the motor turns around 180° or so, in and out to allow or disallow you to pull back on the brake lever. It looks binary, either it's on or off.. My question is.. Why don't you program the motor to turn slower on the way back and also not turn all the way back. It would reduce your delay. Thanks for your great video. Very inspiring.
I was thinking something similar, start fast then ramp down over time to reduce overshoot, both on application and release of the brakes. That or tune the limits of the servo so that the brakes never turn off completely.
Tom your topics for videos are extraordinarily excellent. I am studying automotive engineering and the actual ABS works in a manner very close to what you were trying to achieve. Of course the hardware response time there is ridiculously high. But the theory remains the same. Enjoyed the video. I might just try to make my own abs bike someday. Preferably hydraulic. Also difficult to achieve higher response in cable operated brakes as they have longer travels. Hydraulics on the other hand have negligible travel.
After watching this again, I think the system could benefit greatly by having a force sensor mounted to log tension along the brake cable or hydraulic pressure in the brakes. I'm thinking that a logged force could be used to find a target for the stepper motor during slip events. My thinking is that once the abs activates, the swinging between stopped and rolling can be reduced. Might not be able to prevent the wheel locking the first time, but maybe could prevent the wheel from locking a second time.
Antilock brakes on a mountainbike are pretty pointless but on a motorbike the new 'cornering abs' systems are brilliant. Still very clever stuff though and great video :)
@@Mike-oz4cv just cus in mountainbiking when you're really pushing you are always sliding, i just think it'd have to be a massively over complicated design to actually benefit you and then its got to be fully weather proof, light etc, i just think its a system thats not really needed. It makes for great watching though and im not taking anything away from Tom he amazes me with his ideas
@Kris Roberts yeah exactly, and because of the lack of weight, braking in a straight line very rarely results in a tuck unless you have the reactions of an 80 year old 😄
@@mark675 I agree that for mountain biking, an ABS system is basically useless. However, we are not always sliding when we are really pushing the limits. We just learn to use our brakes. One of the first things we learn about our brakes is how to modulate the power. There is hardly even the need to lock up the wheel. In fact, it's mostly a big no no to lock up your wheel. Over time, you just get used to the feedback and the amount of power needed to brake safely. And of course, the feedback from the brake lever is enough to tell you whether you are locking up the wheel or not.
Andrew Lam your finger is the best abs in mountain biking, especially since there are specific situations where you want to lock it up and this system wouldn't allow that. Having it on your rear wheel would be even worse because skidding helps getting around things quickly and the rear brake doesn't slow you down much at all
@@PraveenKumar-cj4mu I agree Toyota must put in so much wotk into their brakes. There engineers send hours and hours wotking on their abs system. They wotk from dawn to dust. Now if you excuse me, Im abt to go suc thr dck!
Why not use a modified slip ratio equation? SR= (av-aw)/av, using acceleration instead of velocity. That way you don't have to work with a estimated speed but instead values that you know exactly.
The difference it that the velocity of the bike isn't know explicitly, it can only be inferred from the acceleration. In a perfect world that is fine but since we are using an accelerometer to sample the acceleration and it can sample only at a certain frequency there will be skips and so in the measurements, windows of data missing. This will give up a estimate velocity that is close but not exactly the same as the real world velocity. There is also some computational overhead inherent to calculating a value that wouldn't be there in explicit values causing a slight time difference between the inferred bike speed and the measured wheel speed. If all values are measured instead of inferred then this would give the overall most precise value that could possibly obtained. It might not account to much difference honestly but the possible reward (a vast improvement) out weighs the possible risk (negligible at best, 20 seconds extra of dev time to change the code).
@@kkrampus Its the same number, it doesnt matter if there is a additional factor or not, all the uncertainties are the same. The time it takes to calculate this is somewhat irrelevant, unless the µC he uses runs at 10Hz....
Honestly true, but it is still a factor that I would design around. And remember the error in the estimated velocity grows exponentially, as the bike velocity is not just Vo-deltaV its instead Vo-deltaV1-deltaV2-deltaV3.... (at 1kHz which is the max accelerometer refresh rate). If its only .1% out each refresh that could easily become a double digit error within moments. If we instead use fresh acceleration values, values that don't rely on previous values from 1000+ loops ago, there is no possibility for this error. In fact set up like this there is no need for an "activation mode", it can be running all the time as there is no reliance on an original velocity To me its just an error that I don't want to worry about with a simple fix to get around it.
My ABS is "stop pressing the brakes when it locks and press the brakes again, repeat that until the bike stops" It works fine, faster than your mecanism in the video, and it saved me from many crashes. Pd: Sorry if i wrote something bad, I'm still practicing my english.
I did use a bicycle to go to work after 10 years of driving abs-enabled motorbikes. I did end up doing 2 stoppies during emergency breaking, one of which did send my flying over the bars. that's the price for getting used to ABS.
it probably wasn't much better because he had to put a longer lever which gave more torque but less speed so the time it took for the brakes to disengage/engage was longer.
To make that comparison, you'd have to savely come to standstill with a locked frontwheel (as this is the one he implemented it at and also the one you definitely don't want to have locked up). Which shows pretty well, as it's already been said: The purpose is not a shorter braking time, but maintaining control. And with a faster actuator, that would work decently here.
The system is way, way too slow. He really needed to use a hydraulic system to do this, not just because it'd respond faster, but also because it'd let him keep his brake fully engaged with his hand while the pressure is varied like a real abs system.
Atlantic Film a sufficient low-pass filter would need to have a corner frequency of perhaps 1Hz to prevent the system from oscillating, which would put too much of a delay in the entire system.
You hit the nail on the head perfectly when you said who would require an anti lock brake system on a push bike but nevertheless the walk was impressive but if you integrated the system stopping you going over the handlebars that would be very useful good pal.
My bike has rim brakes, and it stops instantly if I squeeze the brakes all the way, at any speed. So, in that case, this system would prevent unintentional front flips.
Themazeful If you only use the rear brake, you can’t stop nearly as quickly. As an experienced cyclist, I’d say I can stop at probably 1/3 to 1/4 the distance when using the front brake vs. the rear.
Gabriel Penner either way you should be using a mixture or the front and rear brakes for maximum braking power. You probably know, but around 80%rear and 20%front braking
I bet human brain is faster than any ABS system, I always use front brake, even on uneven clear ice. And I havent crashed even once. Tires mean everything. How are you supposed to brake in an instant with those? On well suited XC MTB you can hard brake on any survafe without flipping over. Dont see a point of having ABS. If you use front brake and start to lose grip, you let go till you get back on track, it takes maybe 10th of a second. On snow in other hand.... You always have delayed grip which is nearly impossible to predict. Soft braking works well, slowly plowing tire knobs down to ice or ground.
@@mihkus yeah I agree, never had any issues with braking on the front on my XC MTB. But had issues once on wet leafs, had to brake the whole bike tilted to the side, one food down, to not crash the bike. I wonder how an anti lock system would have handled that situation better than me. 🤔
@@mihkus but regarding the human brain: the main difference here is you're expecting to slip in these conditions and slide sideways. If it's dark and there's sand on the road you might be cought by surprise - the human reaction time then increases to up to one second. An electronic system always reacts in an predicted way and timing.
@@mihkus "I bet human brain is faster than any ABS system" Well - you are mistaken in that. assumption. Humans are many orders of magnitude slower. For many things like reacting to a sudden change in the direction while riding your bike your body reacts before your brain does - the spinal cord can do some rudimentary reactions. And still we are talking about taking 2 digit milliseconds at best. The only reason why some ABS systems are slow is cause the system needs to be reliable but not misfire constantly.
Why not go hydraulically actuated powered by an accumulator and pump assembly? Then all you need to do is modulated the pressure output, not wait for the motor to move.
@csalad True. And even if they don't come preloaded, hydraulic brakes can be installed by ourselves too. And speaking of expensive, even $500 dollars is expensive me as a school student from a middle class family.
You could add a value for brake force and combine it with wheel speed to sample the break force right before the wheel skids, so the abs tries its best to avoid locking
this is briliant. Amazing work. I think, like abs on a car, the best way would be to use hydraulics for a quicker response time. However the budget would be immense to develop that
Good job. I would have like to see a side by side comparison of the abs vs without abs to see which stopped quicker. And maybe a little about friction coefficients dynamic vs static on different surfaces.
While mounting the stepper motor like you did is probably the easiest way to do it, I sort of wish that you had mounted it closer to the actual break instead. Also, perhaps it would be possible to attach something onto to the breaking wire. I'm thinking a small clamp that you could attach the stepper motor lever to. Your code is doing its job, but there's a lot of mechanical delays in your design. Since you're using a rope, applying tension to it requires a lot of time, and the stepper motor has to rotate a lot to apply that tension. I know that you're on a budget, but a high torque servo might serve this purpose better than a stepper motor. The horn of the servo would then be attached to the clamp. I think that you'd drastically reduce the mechanical delay this way, and you'd see a breaking action closer resembling that of a car. Apart from that, this concept and execution is probably your most interesting project yet, and that's saying something! Cheers mate.
The solution to the problem of slow response is to use pressure rather than position actuation, fast PWM on that pressure and a duty cycle proportional rather than bang-bang control. To not lock the brakes, don't ever fully brake for a visible length of time. If your response time is not fast enough due to a slew rate problem, don't respond the whole way. Let the braking force back some percent upon slip and then reapply by that percentage the moment you see the next nth optical encoder count. ~One electrical engineer with a control systems background
No idea if you'd ever revisit this project but for hydraulic disc brakes you could possibly plumb in a second lever/cylinder to act as a regulator for the system. This could possibly be servo controlled as the torque required to operate would probably be significantly lower as it doesn't have to operate against your hand. I really love all your projects :)
So using the same knowledge you could create a traction control system for an electric bike like mine. I have a hub motor in my front wheel. If I'm going up a slippery hill and the front wheel loses traction while under power I may go down. If you could sample the speed of both wheels and cut power to the motor when the front wheel is going faster than the back wheel it would prevent some crashes and only require some code and the speed sensors.
What you're talking about is ASR (or simply TCS - Traction Control system), which does exactly what you say - it prevents slip on driven wheels, it's a little more sophisticated than matching wheel speeds, it does allow for some slip, I wonder if any electric bike has that built in.
@@TomaszDominikowski the Alta mxr dirtbike is 80 volt dc driven. 50bhp at the wheel! It has many different "maps" for acceleration. It doesn't just dump all amperage on rear wheel or they woukd spin up to 75mph immediately. Its a progresive step-up system that increases the curve of power as actual road speed increases. Rear always spins but spins just a little faster than road speed. Very interesting stuff. Oh as new maps are written traction can be dynamically changed on the fly as different surfaces are encountered.
I've been thinking of doing that on my off road escooter. I'm thinking using an arduino pro mini to drive a mosfet to PWM the throttle signal wire, with a couple different maps to switch between. Normal, 90%, 80%, 70% acceleration and after a certain time it just goes to on constantly so it passes through. After like 10 seconds or so it will probably have traction, if not chop throttle and program will run again.
@@miker3174 yes the antilock braking on an e-vehicle is based around real time dynamic drive, and braking (drive in reverse). The different aspects of dynamic drive are acceleration, mass, incline, grade, traction, slip ect... All this info in reverse can be used to submit substantially better brake modulation than a human can.
try using a gearbox and a brushless or brushed motor, with some math you'll find that you don't need a very powerful motor so you won't need to spend too much if you calculate right the gear reductions. you can cut the cogs with the cnc cheers!
1:08 "Why does it keep floating out of focu... ooooh there we go, that's why!!!!" 🤣 First video I've seen, and I'm already a big fan haha.. Hooray RUclips/Google hoarding my data, now knowing I recently got a bike 😁🤗
The fact that ABS didn't allow him to lock the front wheel, and understeering all the way to the side and fall, or get thrown over the handle bar is already good.
Couldn't you use the delta of the velocity to determine where the deceleration is heading, then slowly let off the breaks until the deceleration returns to a safe range? This would make the breaking system more fluid rather than jarring.
You can also put a speed indicator on the back tire and when the front tire is going slower then the back it would disengaged the front brake enough to get is closer to the back tires speed
He covers this issue at 11:33. Like he said, the issue with that is if your back tire lifts off the ground, or locks itself (his words), then you would no longer have that reference for the ABS to prevent locking of your front tire.
My method: When riding on hard surfaces, (Asphalt or concrete) I adjust the cable tension to the point where the brake lever can be pulled back against the hand-grip without locking the wheel. Same for softer surfaces, like gravel, dirt or sandy terrain. Just a bit less grip on the calipers by further loosening the tension. I usually do this check before a ride on a given type of terrain. In the event of a panic brake, the front wheel won't lock. I do the same, to a lesser degree on the rear brake as well to prevent the fish-tail effect when the rear locks up. Basically, I adjust the brakes so that lock up is not likely, while still giving good braking for the surface conditions.
For the ABS triggering, you could try combining all of the different triggering mechanisms into one: 1. Zero Wheel Speed 2. Peak Deceleration 3. Accelerometer-estimated Slip Angle. Additionally, you could add a GPS speedometer, which could act as a backup for measuring the bike's speed. You could also add a pair of compasses and 6-axis IMUs (one compass and IMU for each wheel, mounted on the frame and fork respectively), these would be a good future proofing mechanism.
@@petermuller608 for some reason I fogot to mention why I said that, it's possible but it's going to be unreliable with all the dirt in the road interfering with it over time.
15:00 How about using an arm from the actuator to brake handle? It seems that the rope hack added so much stretching that it causes extra lag to the system. Another way would be to use actuator to move *pivot point* of the brake handle instead of moving the handle itself. With adequate movement of the pivot point, the same handle pressure can result in different hydraulic pressure. For example replacing the brake handle pivot bolt with a part that has off-center middle section it could move the handle pivot point a lot with only 180 degree rotation from the actuator (you obviously cannot ever get more than 180 degree range from actuator with this setup, note that you could also get faster on/off cycle with running actuator full speed in one direction).
I ride a lot in places were you would be prone to brake locking but the more experienced you are the better and the good thing about bikes is you feel what the bike is doing knowing how much brake to apply your fingers will do a much better job at it really, it comes natural to me you can hear it and feel it starting to lock up in a car or motorbike yeah its needed you don't feel the things as good in them since there heavier and more sturdy
I think thebiggest factor for the time for anti lock is the mechanism that unlocks the wheels itself. more torque essentially so you can have a faster acting device. I do understand that was out of the budget. You did great work with what you had!
You're working on a lot of projects related to control theory. Have you considered using a hybrid model predictive controller? You can make a non-linear problem piecewise linear and solve the solution space offline. The end result is a lookup table for appropriate linear feedback at a given point in time. Doing the explicit calculation offline should allow you to reduce computation onboard as long as the state space is relative small. That way you can run predictive control algorithms on your cost efficient compute.
What made you go for a stepper motor? A geared DC motor (such as a windshield wiper motor) would have plenty of oomph as well as speed at a very low price
I suggest switching that to rear wheel anti slip by lowering the power to the motor by a ratio of the speed between the front and back wheel, as braking the rear wheel doesn't make the front lift up. You could also setup basically a big optical mouse with open cv by adding a camera or 2 pointing down.
When you mentioned accelerometer, you reminded me of inertial reference units used in aircrafts to measure position of the aircraft. Maybe something similar can be used for predicting speed even if both wheels lock up
Very Cool. Big up man. First improuvement : change the rope for a metal cable, tune the distance of reaction on the cable. Pproposal : it is all about tension of the brake cable, a S shape for lengthen the cable run, with some spring for reduce the engine load. (sorry for my english, i'm a french :c ) :p
I just had an idea for solving the mechanical issue: Compressed air. It would probably only work with rim brakes, but basically, you shoot a little jet of compressed air through a hole in the brake shoe, temporarily lifting the shoe off the rim on a cushion of air. You could modulated the braking much more quickly just using an electronically controlled valve. Of course you'd have to have a small tank on board, but it would probably weigh less than that servo and its battery.
I like how he got to the end of all the test and thought "what's the point?" Just saying you can do that manually by cadence braking, if not better than an electronic system
Hello Tom, one way I would improve the response time is to «limit» the minimum value of the actuator. In other words, you could identify the dead zone at at which the motor does not have an effect on the brakes. Bringing the ABS initial value to the edge of that dead zone would probably cut the reaction time by a big factor. For example, if you know that the brakes are released at 40% of the stepper motor max position, you could set the idle position of the motor to 40%, making the motor release the brakes at 41%, that way, the motor would have to turn 1% to release the brake instead of 40%. It could be a workaround for cutting the response time without breaking your budget. In fact, this would be the equivalent of adding a setpoint to an integrator of a PID controller if that were the case! great video! cheers
can't wait for the hydrolic version of this where there are basically two brake levels that work opposite of each other. Ideally one isn't a lever at all and doesn't need a servo, or maybe just a cam on one.
Hi Tom, great work !! I think everything you do up to the point of operating the brake lever is fenomenal. To make a more effektive, cheaper and simpler way to pull and release the brake lever I would swap the stepper motor for a cheaper regular DC brushed 540 motor and tilt it slightly forwards so the string can pass on the outside of the lever on the motor axel. (Swap to a more reinforced string as well) Then it could just be left spinning in the same direction pulling and releasing the brake lever much faster. Simple engineering without any motor control electronics just a relay to kick it in when all criteria's are met :) The sound would probably also be more similar to a true ABS system as well.
My ABS is called "the brakes aren't strong enough to lock up" and it's never given me a problem
u cannot slow enough then? sorry for my english btw.
Mine is simply called "couch", and/or "couch-lock".
Mine is that i dont have any brakes
i'm kinda surprised how effective this system is, never given me a problem either.
I understand you very well 😔😔. Let's hug
Imagine the neighbours watching him through the window, "yeah Bob hes doing it again, riding up and down his garden skidding on his bike"
🤣😆😄👍
*Tom
@@NathanaelKeller *bob
@@NavJordaan bob!
@@NathanaelKeller Bob?
"DIY" and "braking" are a precarious combination
Well I had to glue the pedal back onto my bike
DIY braking reminded me when I was a child I would brake by putting my foot over the rear wheel. I carved the middle of the soles of all my shoes 🤣
Well, sometimes it ends up in braking bad.
Fun Fact: all prototypes are DIY
...but make for an entertaining video
This was fascinating to watch. The biggest improvement to motorbikes over the last 20 years has been in brakes, and I've often wondered how difficult it would be to create a retrofit kit to fit to older bikes that are still more than fast enough, but stop like a grayhound on a polished floor. One of those things that seems conceptually simple, but evidently complex in practice.
16:03 nice demonstration of abs giving you back control over the bike, balance and steering wise!
My abs is called "dont pull the lever all the way" and it works perfectly fine
Jumper_708 same
Brake and release alternately. Old fashioned cadence braking. Still an acquired skill.
Mine is called rim brakes. It almost doesn't brake in the rain
This has so much potential.
yea they said the same when abs came to motorbikes ...then there is that pedestrian that go in front of you one day x(
Why even in commercial devices the brake is released so significantly? The best choice would be a PID control to release slightly the brake to get the best possible deceleration according to tire friction, the deceleration would appear as constant (the PID loopback can be done thousand times a second).
It is not released that much in commercial systems. In a car you have three states build up pressure, hold pressure and release pressure. The abs changes between these cycles 15-20 times per second.
In most situations this is more than enough.
@@NathanaelKeller My daily driver has one of the oldest antilock systems Ford put in a commercial vehicle, and it's very slow at just above 4 cycles per second. From experience, I can tell you that it's plenty good enough to keep traction even in heavy snows, even if it's perhaps just a bit too aggressive, increasing stop distance just a bit. It does sound like an old, heavy machine gun though, so that's cool.
Absolutely the first thing which came to mind, PID or PD. Best way to decrease overshoot and minimize rise time since the wheel speed and acceleration is changing dynamically with the surface contact. Completely unnecessary on a bicycle, but an enjoyable design process.
@@Nolan001001 the trouble with a PID for this specific situation is it needs to be highly tuned which is possible if you have consistent parameters, but an ABS system needs to work across a wide range of grip levels, so it can't be tuned easily. In reality, the software behind real ABS systems is highly sophisticated taking in multiple sensor inputs and determining what is the maximum grip that can be achieved. A lot of modern systems also now include active stability control which will brake the vehicle unevenly left-right and front-back to try to help prevent spins.
@@lloydmorrissey Very true, I was mostly speaking related to the project in the video itself, how to get this specific application to work "better", like I said previously, would be fun to do. And besides I'm not a control systems engineer, I wouldn't even begin to fathom the block diagram of a true abs system and the resulting transfer functions.
Physics is the greatest teacher of brake control on a mountain bike. Crash enough times and you'll figure out what you're doing wrong to fix it. Ideally, you'll be a one trial learner.
Yes, I have gone over the handle bars enough to figure this out. I am obviously not a one trail learner, my bike knows
Skid the front wheel for 0.2 seconds you'll be on the deck, too - for the system to work it would have to somehow prevent any skid from even starting...
I was going at 35km/h. Suddenly a person came in front. I have rim brakes and very bad tyres. And also don't have a front brake. There was little space from the side so couldn't go in speed. Did full emergency brakes tyres skid and turned to the left, then right and then left. What a drift. Was so thrilling but wasn't intentional.
I learned everything the hard way as a kid so I don't have to now.
@@gill_rides7618yep, front brakes are to be used gently, rear brakes can be locked unless it’s wet
Wet grass? In England? Where did you find it?
😂😂
omniversling everywhere
Lazy Wolf wooosh
imported from France.
nice one
The motor response is too slow, i have a motorcycle with abs and its abs is phenomenon save me couple of times in emergency braking
because that system is on a different league and works directly on the brakes
Shut up
you are a smarty pants aren’t you...
@@RageShredder if you don't like facet, i can't help you either
Yeah I just now threw in a comment about how he needs a huge motor because he's trying to counter-act his hand, and he's trying to do it at the handle instead of the brakes. With the right mechanism, you could use a solenoid instead of a servo, and have response times in the 10s of milliseconds.
This was far more interesting and technical than I first expected. Loved it! From a control systems engineer.
Michael Steeves hey vsauce man
@@alice11307 Not even Close
Rather than having an ABS system that releases when it senses a locked wheel, would it work to make them completely fly-by-wire where the brake lever is just an input to the computer to indicate percentage of braking force? At 100%, the computer would only apply the maximum force that would still keep the braking of the wheel within the ideal slip ratio portion of the graph. So, instead of trying to modulate the pressure after it's locked, only enough braking force would be applied to maximize deceleration without the lock up.
Good idea, yet has a safety concern, if your battery runs out, you wont be brakin mate....
Also more difficult to manufacture than a pressure release and repump system
@@danieltabrizian Haha, true that, but the brakes won't lock up. :-D
Great idea but after all it would cost a fortune to invest in such a system for a avarage commuter
Big problem - 100% braking depends on the road surface, grease,ice,wet rims,etc. So I can't see how you could design the system.
This is just fantastic. This video had me thinking about all sorts of things including why the ABS judders your wheels in an emergency braking situation.
The juddering is mainly noticible in older systems that could only cycle between full clamp and full release. Repeat this nearly 100 times per second and you induce some fun vibrations.
I thought this was going to be a video of you 3D printing your own bike out of ABS
ikr, he should have
what a wasted opportunity
Your data should also allow you to calculate the friction of the surface from the first lock. You can then use this to predict the maximum safe breaking force and release the break to the ideal position. You'd just need to adjust the break pressure curve to allow the wheel to speed up again before clamping to the ideal pressure.
Spray some wd40 on the brakes then they wont lock 🤣🤣
Alec Parkin I was greasing my hub, didn’t realise I got some on my disk, tested me hub and this used by breaks to slow only to realise I have got grease in my pads
NxvaSix every time I go out in the snow, my disks get contaminated... so annoying to fix
@@nxvasix8696 I have the same problem, any idea on how to fix it?
Oil on brakes will make them squeal.
@@dewexdewex wd40 is literally the opposite of oil. The whole point of it is to destroy oil
16:51 i think he broke his leg
Vugh de Ziestem LMAO
BRUH
Holy shit
He should have mentioned it but that cracking sound though
@@eamonlee159 that sound is the bike hitting the ground
Well done mate! It really shows how hard is to replicate such easy to understand mechanisms. The first brake release mechanisms looked way cooler though ;-)
Lol am I the first reply
Heart this comment plz
Upgrade the stepper :)!!!!
@@samuelyoung2671 or a smaller stepper, and hydrolic brakes. ... and the 120slits cut into the brake rotor itself. . and an AI camera system too create a speed differential, - and of course, do it on both wheels. ;-D. totally fantastic tom !
@Lassi Kinnunen its really quite exciting, for a push bike, - and i thought an electric motor was cool.... wtf. - engineering pwns * checkout my "free energy motor* demonstration ruclips.net/video/sJ3zjje8hgw/видео.html
Alternatively, you could just modulate the brake and have the same result
Very cool work! Perhaps a solenoid or other linear actuator would have better mechanical response time than the motor.
Very good points about real-world applications too
Wait...youre here...btw big fan of your ksp vid but....very cool seeing you here
I think the reason automotive ABS uses hydraulics is the immediacy provided by incompressible fluids. The exact moment the actuator is initiated the effect is applied. Practically zero latency.
Long before ABS, as a professional emergency services driver (fire) we were taught cadence braking; essentially manually operated ABS. I used that when driving my own car pre-abs and do the same with the front brake on both motorcycle and bicycle. It is an acquired skill but still very effective. It works best on hydraulic cycle brakes but I have also found that the front disc is so easily contaminated with road muck that it effectively turns itself into ABS by alternately slipping and locking but it is very noisy! I used to clean my front disc and pads very regularly but found it locked up on gravel far too easily so I left it alone. It works!!!! It's noisy but very effective (self) cadence braking. No need for expensive ABS as long as you can stand the noise plus the noise itself turns you into a very careful breaker; effectively a verbal cadence braking warning. It works for me and has done for many years. I'm 71 years of age and still riding 100+ miles per week safely. 😀
Tom, how about using a solenoid for the actuation? The response times would be very quick
cheap, too.... Huh, now I want to test that on my bike.
I would say reverse the system - pre-tension the brake so the brake line is held normally but the solenoid can slacken it very fast
I was looking at the mechanism and thought servo. They appear to have fast response and good torque
@@Gunzee honestly anything that wasn't a servo motor would have been 100x better. There was no need here for precise control (he connected the thing with a piece of rope)
@@zaprodk good idea
Glad you ditched the first prototype, as you were running into the prpblem with this being a cable based abs braking system.
When you look at a cars abs system, it uses hydraulic brakes and a very expensive abs pump to quickly and precisely pump individual brake circuits.
You want the fastest possible mechanism controlling the pulse and release part, like a solenoid or high end RC servo. These slow stepper motors arent cutting
This comment is 100% correct same thing with motorcycle they also have hydrolic system as well thumb up 👍
@@beckyg4860 *hydraulic
The Bosch E Bike ABS also uses a hydraulic valve to engage and disengage the brake, they just don't use a pump feeding back the hydraulic fluid, you have to do this manually with the lever.
Love this kind of experiments, it takes a lot of time though, but research is worth it.
Joyplanes RC ya ves
"Maybe third time's the charm?" - unnamed ABS manufacturer
I mean.. there are four of them on a car so number 3 and 4 working is 50%. You'll need to have an upgrade path for the customer.
That's one of the videos that made me retrofit an ABS on my bike (which never had it as an option from factory). Oh and later upgraded it into a traction control system with 4 different modes. And it works like charm - saved my ass a couple of times. Thanks for an inspiration!
I was getting concerned when half way through the video you hadn't mentioned accelerometers yet.... You were just trolling us though!
thats not trolling, thats storytelling :D
Accelerometers are useless, theres no way to know between a locked wheel and you breaking slowly. The only way would be if you correlated the accelerometer to the wheel decceleration, and activated the ABS if they get too far away.
Older ABS systems did not use accelerometers. The really early systems were entirely mechanical, have at look for the "Lucas Girling Stop Control System", a very clever mechanism.
the xiaomi m365 use e abs sooo why not.
@@laharl2k not if you put the accelerometer on the wheel itself.
Are you sure the motorbike uses an optical sensor for the wheel speed? I've only ever seen vehicles using hall effect sensors, since the optical sensors would be subject to dirt and grime blocking the light emitter or detector. I've been looking forward to the next video, so glad it came out! Another fine video presentation. Nicely done!
Lol
Yeah, the MC is hall effect sensor (edit: Induction sensor actually). Never seen one done with optical. Car or MC. Or anti skid on aircraft for that matter. That's where it was original developed btw
I wasn't 100% sure, but just assumed as it used a slotted disk. I would have thought a hall effect sensor would have a solid disk with the magnets embedded inside?
@@TomStantonEngineering It is not even hall effect sensor. The slotted wheel is called a reluctor ring and it induces a current in the sensing coil (in the sensor).
Gotcha, that makes sense! I suppose it still achieves the same result, just using a different sensor
Dude, you have one of the best hobbies ever! I wish I could come up with stuff like that by just tinkering around.
I really appreciate your analysis of *whether* and *for whom* ABS might actually make sense on bicycles. Requirements engineering is a key part of engineering!
You're actually designing a PID controller. Very educative. Thumbs up!
You could use a servo controlled pressure unloading cylinder ,just put it in line with the caliper and move it in and out with a servo....
This is a great idea. Would be much faster response
i was thinking along the same lines thinking linear actuator opening a chamber like a piston but it would have to be small enough strong enough and quick enough
@@oldskoolhead0 Yeah holding force might be a problem ,but I think a piston on a linear actuator would be quite hard to move. Especially if you kept the cross section down to limit the force that could be applied...
@@swamisamantharella i suppose it could be put on a cantilever like the brake lever itself or maybe even put on the end of the brake lever between the handlebars (which would also mean its normally closed instead of normally open, so no need to hold anything for too long either) and it would only have to fight against your hand
i thought so. this guy is stupid, tryin with wired brakes...
Hi Tom. I see that the motor turns around 180° or so, in and out to allow or disallow you to pull back on the brake lever. It looks binary, either it's on or off.. My question is.. Why don't you program the motor to turn slower on the way back and also not turn all the way back. It would reduce your delay. Thanks for your great video. Very inspiring.
I was thinking something similar, start fast then ramp down over time to reduce overshoot, both on application and release of the brakes. That or tune the limits of the servo so that the brakes never turn off completely.
Because this is just some arduino slapped-together piece of code, not even close to something properly engineered.
bro this guy got so much in his head
Fr
Nerd keeps Nerding..
Keep on Nerding out for all of us Tom.
The Dave lol
Sorry but this is very basic, and it was a fail.
He needs to go out and get laid.
Tom your topics for videos are extraordinarily excellent. I am studying automotive engineering and the actual ABS works in a manner very close to what you were trying to achieve. Of course the hardware response time there is ridiculously high. But the theory remains the same. Enjoyed the video. I might just try to make my own abs bike someday. Preferably hydraulic.
Also difficult to achieve higher response in cable operated brakes as they have longer travels. Hydraulics on the other hand have negligible travel.
After watching this again, I think the system could benefit greatly by having a force sensor mounted to log tension along the brake cable or hydraulic pressure in the brakes. I'm thinking that a logged force could be used to find a target for the stepper motor during slip events. My thinking is that once the abs activates, the swinging between stopped and rolling can be reduced. Might not be able to prevent the wheel locking the first time, but maybe could prevent the wheel from locking a second time.
Antilock brakes on a mountainbike are pretty pointless but on a motorbike the new 'cornering abs' systems are brilliant.
Still very clever stuff though and great video :)
Why pointless?
@@Mike-oz4cv just cus in mountainbiking when you're really pushing you are always sliding, i just think it'd have to be a massively over complicated design to actually benefit you and then its got to be fully weather proof, light etc, i just think its a system thats not really needed.
It makes for great watching though and im not taking anything away from Tom he amazes me with his ideas
@Kris Roberts yeah exactly, and because of the lack of weight, braking in a straight line very rarely results in a tuck unless you have the reactions of an 80 year old 😄
@@mark675 I agree that for mountain biking, an ABS system is basically useless. However, we are not always sliding when we are really pushing the limits. We just learn to use our brakes. One of the first things we learn about our brakes is how to modulate the power. There is hardly even the need to lock up the wheel. In fact, it's mostly a big no no to lock up your wheel. Over time, you just get used to the feedback and the amount of power needed to brake safely. And of course, the feedback from the brake lever is enough to tell you whether you are locking up the wheel or not.
Andrew Lam your finger is the best abs in mountain biking, especially since there are specific situations where you want to lock it up and this system wouldn't allow that. Having it on your rear wheel would be even worse because skidding helps getting around things quickly and the rear brake doesn't slow you down much at all
Superb job man, I'm greatly impressed with the amount of work that went into this one! Looking forward to the next project. :)
el abs es para putos, tanto en autos, motos o bicicletas...sin excepcion
@@PraveenKumar-cj4mu I agree Toyota must put in so much wotk into their brakes. There engineers send hours and hours wotking on their abs system. They wotk from dawn to dust. Now if you excuse me, Im abt to go suc thr dck!
@@Kaosad69 please do
Why not use a modified slip ratio equation? SR= (av-aw)/av, using acceleration instead of velocity. That way you don't have to work with a estimated speed but instead values that you know exactly.
I thought the same
Since you are using the same number... what would change? In both cases its just the accuracy of the sensor that matters (or limits the system).
The difference it that the velocity of the bike isn't know explicitly, it can only be inferred from the acceleration. In a perfect world that is fine but since we are using an accelerometer to sample the acceleration and it can sample only at a certain frequency there will be skips and so in the measurements, windows of data missing. This will give up a estimate velocity that is close but not exactly the same as the real world velocity. There is also some computational overhead inherent to calculating a value that wouldn't be there in explicit values causing a slight time difference between the inferred bike speed and the measured wheel speed.
If all values are measured instead of inferred then this would give the overall most precise value that could possibly obtained. It might not account to much difference honestly but the possible reward (a vast improvement) out weighs the possible risk (negligible at best, 20 seconds extra of dev time to change the code).
@@kkrampus Its the same number, it doesnt matter if there is a additional factor or not, all the uncertainties are the same. The time it takes to calculate this is somewhat irrelevant, unless the µC he uses runs at 10Hz....
Honestly true, but it is still a factor that I would design around. And remember the error in the estimated velocity grows exponentially, as the bike velocity is not just Vo-deltaV its instead Vo-deltaV1-deltaV2-deltaV3.... (at 1kHz which is the max accelerometer refresh rate). If its only .1% out each refresh that could easily become a double digit error within moments.
If we instead use fresh acceleration values, values that don't rely on previous values from 1000+ loops ago, there is no possibility for this error. In fact set up like this there is no need for an "activation mode", it can be running all the time as there is no reliance on an original velocity
To me its just an error that I don't want to worry about with a simple fix to get around it.
My ABS is "stop pressing the brakes when it locks and press the brakes again, repeat that until the bike stops" It works fine, faster than your mecanism in the video, and it saved me from many crashes.
Pd: Sorry if i wrote something bad, I'm still practicing my english.
This has been on my mind for awhile, I am subscribing to watch your journey!
I did use a bicycle to go to work after 10 years of driving abs-enabled motorbikes. I did end up doing 2 stoppies during emergency breaking, one of which did send my flying over the bars. that's the price for getting used to ABS.
I was waiting for a proper comparison of break time/distance with and without the "ABS" !
The purpose of abs is more to maintain control. It actually can result in slightly longer braking distances in order to maintain control.
@@Redbikemaster I'm sure it also helps to maintain static friction rather than the much smaller dynamic friction.
it probably wasn't much better because he had to put a longer lever which gave more torque but less speed so the time it took for the brakes to disengage/engage was longer.
@@mannyr9225 it probably doesn't, seeing it's a binary on/off system with no fine adjustment.
To make that comparison, you'd have to savely come to standstill with a locked frontwheel (as this is the one he implemented it at and also the one you definitely don't want to have locked up). Which shows pretty well, as it's already been said: The purpose is not a shorter braking time, but maintaining control. And with a faster actuator, that would work decently here.
Why not release the brake only partially/ using PID to match the wheel speed to little below the speed calculated by the accelerometer?
His system has far too much hysteresis to do that properly.
The system is way, way too slow.
He really needed to use a hydraulic system to do this, not just because it'd respond faster, but also because it'd let him keep his brake fully engaged with his hand while the pressure is varied like a real abs system.
@@Scrogan valid argument but imo still a better solution then the current. Also with filtering it would improve
Atlantic Film a sufficient low-pass filter would need to have a corner frequency of perhaps 1Hz to prevent the system from oscillating, which would put too much of a delay in the entire system.
Cars do this
You hit the nail on the head perfectly when you said who would require an anti lock brake system on a push bike but nevertheless the walk was impressive but if you integrated the system stopping you going over the handlebars that would be very useful good pal.
My bike has rim brakes, and it stops instantly if I squeeze the brakes all the way, at any speed. So, in that case, this system would prevent unintentional front flips.
If you brake the back wheel you shouldn't flip.
Themazeful
If you only use the rear brake, you can’t stop nearly as quickly. As an experienced cyclist, I’d say I can stop at probably 1/3 to 1/4 the distance when using the front brake vs. the rear.
@@21Trainman slow down a little maybe?
Gabriel Penner either way you should be using a mixture or the front and rear brakes for maximum braking power. You probably know, but around 80%rear and 20%front braking
@@kauvik_ it keeps the bike from tipping over when barking is distributed instead of putting on the load on one point.
16:08 actually pretty impressive if you realize how fast the human brain is responding to skidding with a steering input to hold the balance 😮
I bet human brain is faster than any ABS system, I always use front brake, even on uneven clear ice. And I havent crashed even once. Tires mean everything.
How are you supposed to brake in an instant with those?
On well suited XC MTB you can hard brake on any survafe without flipping over.
Dont see a point of having ABS.
If you use front brake and start to lose grip, you let go till you get back on track, it takes maybe 10th of a second.
On snow in other hand.... You always have delayed grip which is nearly impossible to predict. Soft braking works well, slowly plowing tire knobs down to ice or ground.
@@mihkus yeah I agree, never had any issues with braking on the front on my XC MTB.
But had issues once on wet leafs, had to brake the whole bike tilted to the side, one food down, to not crash the bike. I wonder how an anti lock system would have handled that situation better than me. 🤔
@@mihkus but regarding the human brain: the main difference here is you're expecting to slip in these conditions and slide sideways. If it's dark and there's sand on the road you might be cought by surprise - the human reaction time then increases to up to one second.
An electronic system always reacts in an predicted way and timing.
@@mihkus "I bet human brain is faster than any ABS system"
Well - you are mistaken in that. assumption. Humans are many orders of magnitude slower.
For many things like reacting to a sudden change in the direction while riding your bike your body reacts before your brain does - the spinal cord can do some rudimentary reactions. And still we are talking about taking 2 digit milliseconds at best.
The only reason why some ABS systems are slow is cause the system needs to be reliable but not misfire constantly.
@@mihkus Hot a sleepy texting young moms brain, that is concerned with a baby screaming in the passenger seat.
Why not go hydraulically actuated powered by an accumulator and pump assembly? Then all you need to do is modulated the pressure output, not wait for the motor to move.
Would be the proper solution, but the answer is budget.
Budget and the brakes on the bike are powered with a cable, not hydraulics like a car or a motorcycle
Shedman well expensive bikes do come with hydraulic brakes 🙏
@@suryagurung7793 i guess it depends how you define "expensive", but hydraulic brakes can be had on a bicycle around the 500 USD range.
@csalad True. And even if they don't come preloaded, hydraulic brakes can be installed by ourselves too.
And speaking of expensive, even $500 dollars is expensive me as a school student from a middle class family.
You could add a value for brake force and combine it with wheel speed to sample the break force right before the wheel skids, so the abs tries its best to avoid locking
this is briliant. Amazing work. I think, like abs on a car, the best way would be to use hydraulics for a quicker response time. However the budget would be immense to develop that
Good job. I would have like to see a side by side comparison of the abs vs without abs to see which stopped quicker. And maybe a little about friction coefficients dynamic vs static on different surfaces.
While mounting the stepper motor like you did is probably the easiest way to do it, I sort of wish that you had mounted it closer to the actual break instead. Also, perhaps it would be possible to attach something onto to the breaking wire. I'm thinking a small clamp that you could attach the stepper motor lever to. Your code is doing its job, but there's a lot of mechanical delays in your design. Since you're using a rope, applying tension to it requires a lot of time, and the stepper motor has to rotate a lot to apply that tension. I know that you're on a budget, but a high torque servo might serve this purpose better than a stepper motor. The horn of the servo would then be attached to the clamp. I think that you'd drastically reduce the mechanical delay this way, and you'd see a breaking action closer resembling that of a car. Apart from that, this concept and execution is probably your most interesting project yet, and that's saying something! Cheers mate.
The solution to the problem of slow response is to use pressure rather than position actuation, fast PWM on that pressure and a duty cycle proportional rather than bang-bang control. To not lock the brakes, don't ever fully brake for a visible length of time. If your response time is not fast enough due to a slew rate problem, don't respond the whole way. Let the braking force back some percent upon slip and then reapply by that percentage the moment you see the next nth optical encoder count. ~One electrical engineer with a control systems background
I wish you'd show controls without abs so we can see how much the abs improved braking distance
As an auto mechanic, I finally felt fairly intelligent watching one of your videoes.
No idea if you'd ever revisit this project but for hydraulic disc brakes you could possibly plumb in a second lever/cylinder to act as a regulator for the system. This could possibly be servo controlled as the torque required to operate would probably be significantly lower as it doesn't have to operate against your hand. I really love all your projects :)
This is also how more the more modern ABS systems on some new mountain bikes work ;)
So using the same knowledge you could create a traction control system for an electric bike like mine. I have a hub motor in my front wheel. If I'm going up a slippery hill and the front wheel loses traction while under power I may go down. If you could sample the speed of both wheels and cut power to the motor when the front wheel is going faster than the back wheel it would prevent some crashes and only require some code and the speed sensors.
What you're talking about is ASR (or simply TCS - Traction Control system), which does exactly what you say - it prevents slip on driven wheels, it's a little more sophisticated than matching wheel speeds, it does allow for some slip, I wonder if any electric bike has that built in.
@@TomaszDominikowski the Alta mxr dirtbike is 80 volt dc driven. 50bhp at the wheel! It has many different "maps" for acceleration. It doesn't just dump all amperage on rear wheel or they woukd spin up to 75mph immediately. Its a progresive step-up system that increases the curve of power as actual road speed increases. Rear always spins but spins just a little faster than road speed. Very interesting stuff. Oh as new maps are written traction can be dynamically changed on the fly as different surfaces are encountered.
I've been thinking of doing that on my off road escooter. I'm thinking using an arduino pro mini to drive a mosfet to PWM the throttle signal wire, with a couple different maps to switch between. Normal, 90%, 80%, 70% acceleration and after a certain time it just goes to on constantly so it passes through. After like 10 seconds or so it will probably have traction, if not chop throttle and program will run again.
@@miker3174 yes the antilock braking on an e-vehicle is based around real time dynamic drive, and braking (drive in reverse). The different aspects of dynamic drive are acceleration, mass, incline, grade, traction, slip ect... All this info in reverse can be used to submit substantially better brake modulation than a human can.
try using a gearbox and a brushless or brushed motor, with some math you'll find that you don't need a very powerful motor so you won't need to spend too much if you calculate right the gear reductions. you can cut the cogs with the cnc
cheers!
i've had to emergency break a few times on my bike and i've skidded on ice a few times too but i've been biking daily for 4 years
1:08 "Why does it keep floating out of focu... ooooh there we go, that's why!!!!" 🤣
First video I've seen, and I'm already a big fan haha..
Hooray RUclips/Google hoarding my data, now knowing I recently got a bike 😁🤗
The fact that ABS didn't allow him to lock the front wheel, and understeering all the way to the side and fall, or get thrown over the handle bar is already good.
Couldn't you use the delta of the velocity to determine where the deceleration is heading, then slowly let off the breaks until the deceleration returns to a safe range? This would make the breaking system more fluid rather than jarring.
AFAIK that's how proper ABS works.
This wouldnt work without pre programming for the surface and weather conditions
You can also put a speed indicator on the back tire and when the front tire is going slower then the back it would disengaged the front brake enough to get is closer to the back tires speed
He covers this issue at 11:33. Like he said, the issue with that is if your back tire lifts off the ground, or locks itself (his words), then you would no longer have that reference for the ABS to prevent locking of your front tire.
1:17 I cracked up when I saw that
My method: When riding on hard surfaces, (Asphalt or concrete) I adjust the cable tension to the point where the brake lever can be pulled back against the hand-grip without locking the wheel. Same for softer surfaces, like gravel, dirt or sandy terrain. Just a bit less grip on the calipers by further loosening the tension. I usually do this check before a ride on a given type of terrain. In the event of a panic brake, the front wheel won't lock. I do the same, to a lesser degree on the rear brake as well to prevent the fish-tail effect when the rear locks up.
Basically, I adjust the brakes so that lock up is not likely, while still giving good braking for the surface conditions.
Uhh how do I do that with hydraulic disc brakes?
For the ABS triggering, you could try combining all of the different triggering mechanisms into one:
1. Zero Wheel Speed
2. Peak Deceleration
3. Accelerometer-estimated Slip Angle.
Additionally, you could add a GPS speedometer, which could act as a backup for measuring the bike's speed.
You could also add a pair of compasses and 6-axis IMUs (one compass and IMU for each wheel, mounted on the frame and fork respectively), these would be a good future proofing mechanism.
Hey Tom, it's quite impossible for a commercial motorbike to use an optical sensor for wheel speed, that must be an inductive sensor. Keep it up!
Why?
@@petermuller608 for some reason I fogot to mention why I said that, it's possible but it's going to be unreliable with all the dirt in the road interfering with it over time.
@@youssefaly97 thanks for the explanation!
15:00 How about using an arm from the actuator to brake handle? It seems that the rope hack added so much stretching that it causes extra lag to the system.
Another way would be to use actuator to move *pivot point* of the brake handle instead of moving the handle itself. With adequate movement of the pivot point, the same handle pressure can result in different hydraulic pressure. For example replacing the brake handle pivot bolt with a part that has off-center middle section it could move the handle pivot point a lot with only 180 degree rotation from the actuator (you obviously cannot ever get more than 180 degree range from actuator with this setup, note that you could also get faster on/off cycle with running actuator full speed in one direction).
Next time would be good if you walked through the code on the next one of your videos like this.
Can you do traction/ ESP on a remote control car??
Wrecking the lawn in the name of science. Love it, we can really appreciate the sacrifice! :D
😆haaahaha a new way an keeping the grass down !!.
So now I want to see a hub motor on each wheel, traction control and abs ;)
It probably need a PID controller to prevent oscillations of breaking force.
cool concept but my fingers are the best way of stopping my brakes from locking up on my MTB.
As soon as the wheel speed drops below the vehicle speed, fire reverse thrusters!
I ride a lot in places were you would be prone to brake locking but the more experienced you are the better and the good thing about bikes is you feel what the bike is doing knowing how much brake to apply your fingers will do a much better job at it really, it comes natural to me you can hear it and feel it starting to lock up in a car or motorbike yeah its needed you don't feel the things as good in them since there heavier and more sturdy
This guy just ruined his backyard to teach us stuff.
Respect.
solenoids would work better than stepper motors.
They would have less power
That is very interesting, but bike not need ABS, because rider can feel brakes
My abs work most of the time, it’s called full suspension and it’s saved me a couple of times from going over the handle bars.
I think thebiggest factor for the time for anti lock is the mechanism that unlocks the wheels itself. more torque essentially so you can have a faster acting device. I do understand that was out of the budget. You did great work with what you had!
You're working on a lot of projects related to control theory. Have you considered using a hybrid model predictive controller? You can make a non-linear problem piecewise linear and solve the solution space offline. The end result is a lookup table for appropriate linear feedback at a given point in time. Doing the explicit calculation offline should allow you to reduce computation onboard as long as the state space is relative small. That way you can run predictive control algorithms on your cost efficient compute.
What made you go for a stepper motor? A geared DC motor (such as a windshield wiper motor) would have plenty of oomph as well as speed at a very low price
Intermittent... lol
>"DIY"
>*uses cnc and 3d printer before the first 2 minutes*
3d printers are definitely common DIY tools these days. CNC routers slightly less so but still not rare, or particularly expensive.
Where in diy does it say cheap?
He's still making it himself
me: *hehehe bike go skrttt skrrrt*
Tom:
*skkkrtttt OOWWWW*
I suggest switching that to rear wheel anti slip by lowering the power to the motor by a ratio of the speed between the front and back wheel, as braking the rear wheel doesn't make the front lift up.
You could also setup basically a big optical mouse with open cv by adding a camera or 2 pointing down.
@@deepmantomar3975 i dont think you were meant to reply to my comment...
Impressive that you were able to make something so simple and useful so complex and dangerous
When you mentioned accelerometer, you reminded me of inertial reference units used in aircrafts to measure position of the aircraft. Maybe something similar can be used for predicting speed even if both wheels lock up
My bike could do ABS, with my rear wheel jumping up and down
My semi truck does that with its trailer when empty.
Cheap abs "using oil on braking part "
*extremely dangerous * *may be fatal*
Can't live life properly without a little risk now can you lol
@@Runedragonx 😁😁
Bruh, just wear off the break pads! Much easier!
just dont use brakes. benefit from the weight reduction anyway
My bike only has front brakes because the rear ones got wet and stopped working somehow even when dry
Very Cool. Big up man.
First improuvement : change the rope for a metal cable, tune the distance of reaction on the cable.
Pproposal : it is all about tension of the brake cable, a S shape for lengthen the cable run, with some spring for reduce the engine load.
(sorry for my english, i'm a french :c ) :p
I just had an idea for solving the mechanical issue: Compressed air.
It would probably only work with rim brakes, but basically, you shoot a little jet of compressed air through a hole in the brake shoe, temporarily lifting the shoe off the rim on a cushion of air. You could modulated the braking much more quickly just using an electronically controlled valve. Of course you'd have to have a small tank on board, but it would probably weigh less than that servo and its battery.
I like how he got to the end of all the test and thought "what's the point?"
Just saying you can do that manually by cadence braking, if not better than an electronic system
I don't think you can do better than an optimised system
1:18 "That autofocus THO"
This makes me feel like a caveman.
Hello Tom, one way I would improve the response time is to «limit» the minimum value of the actuator. In other words, you could identify the dead zone at at which the motor does not have an effect on the brakes. Bringing the ABS initial value to the edge of that dead zone would probably cut the reaction time by a big factor. For example, if you know that the brakes are released at 40% of the stepper motor max position, you could set the idle position of the motor to 40%, making the motor release the brakes at 41%, that way, the motor would have to turn 1% to release the brake instead of 40%. It could be a workaround for cutting the response time without breaking your budget. In fact, this would be the equivalent of adding a setpoint to an integrator of a PID controller if that were the case! great video! cheers
Realistically on a bike it would be much more efficient to just gain the skills to not lock up the wheel, but this is just super cool
He does realize that he could just not apply as much brake right?
poor grass :(
Yeah someone's going to go didlow when they see it!
Colab with GMBN?
William the butchers son productions nah chief ABS had no place in mountain biking, maybe cross country but not in freeride.
can't wait for the hydrolic version of this where there are basically two brake levels that work opposite of each other. Ideally one isn't a lever at all and doesn't need a servo, or maybe just a cam on one.
Hi Tom, great work !!
I think everything you do up to the point of operating the brake lever is fenomenal. To make a more effektive, cheaper and simpler way to pull and release the brake lever I would swap the stepper motor for a cheaper regular DC brushed 540 motor and tilt it slightly forwards so the string can pass on the outside of the lever on the motor axel. (Swap to a more reinforced string as well) Then it could just be left spinning in the same direction pulling and releasing the brake lever much faster. Simple engineering without any motor control electronics just a relay to kick it in when all criteria's are met :) The sound would probably also be more similar to a true ABS system as well.