Calling bike physics/tech experts!

[bogan]

So I'm nearing completion of my electric bike project (electric bucket hopefully) so starting floating ideas for the next big project. Been thinking bout doing an autonomous bike project. Something like this

<object width="480" height="385"><param name="movie" value="http://www.youtube.com/v/XOgkNh_IPjU&hl=en_US&fs=1&"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube.com/v/XOgkNh_IPjU&hl=en_US&fs=1&" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width="480" height="385"></embed></object>

but better with a bit of luck :D

First step will be to adapt my learner bike (VT250) for remote control use. I reckon clutch, throttle, brake, will be easy enough (from theoretical perspective) to sort but steering will somewhat more difficult. It appears that the above guy uses a motor and encoder to steer, which appears to give a rather jerky steering system. I have different ideas, but would like to hear what you guys reckon first?

[Headbanger]

I have different ideas, but would like to hear what you guys reckon first?


I reckon you're a bit mad:shit:

[munster]

Looks interesting. Keen to know more about your electric bucket racer. I've been thinking about something similar for awhile. What sort of motor did you use (voltage, amps etc) and power controller?

[Urano]

i'm on an "almost" electric project too...
pretty interested in that, but maybe we're a bit out of topic.
i'd really like to keep in touch about that, anyway...

[Jonathan]

I reckon you can get inspiration from the fat armless Chinese guy.

<img src="http://neatorama.cachefly.net/images/2009-07/motorcycle-no-hands.jpg">

All you need is a weight in the centre of the bike that leans to the left and right and is connected to the bars just as Mr Yeung there. He doesn't need to push, doesn't need to pull, just leans from side to side.

Guess it is just the same concept as above though...

[bogan]

To those enquiring about the electric, I'll put up a thread once thats done, hopefully it'll turn into a big one so all that have enquired will notice it.

Interesting thought Jonathan, though is the shifting weight important? as steering will shift the CM to start/adjust a turn anyway.

[CookMySock]

The problem with pushing on bars is, the push has to be a pressure not a displacement, so the task is either to engineer the servo to be a force system and then figure out the damping (the hard way IMO), or else use a displacement servo and get some torque feedback and do it all in software.

Steve

[bogan]

The problem with pushing on bars is, the push has to be a pressure not a displacement, so the task is either to engineer the servo to be a force system and then figure out the damping (the hard way IMO), or else use a displacement servo and get some torque feedback and do it all in software.

Steve

Exactly what I thought, however theres a third option, use a voice-coil actuator and position encoder

[R-Soul]

The problem with pushing on bars is, the push has to be a pressure not a displacement, so the task is either to engineer the servo to be a force system and then figure out the damping (the hard way IMO), or else use a displacement servo and get some torque feedback and do it all in software.

Steve

Why do you say it must be a pressure and not a displacement. (and I stand corrected on this but )If you are putting pressure on the bars, but not getting any displacement out of it, you are not going to be making a difference to the steering geometry or lean angle, and hence no difference to the steering itself?

[R-Soul]

How about a four wheeled bike like the 4by or the Yamaha prototype bike?

[bogan]

Why do you say it must be a pressure and not a displacement. (and I stand corrected on this but )If you are putting pressure on the bars, but not getting any displacement out of it, you are not going to be making a difference to the steering geometry or lean angle, and hence no difference to the steering itself?

think what he means (and what I mean), is control the pressure, not the displacement, most motors control displacement rather than pressure. The reason is to allow the bars to move about a little bit with bumps etc, and to avoid big forces that would be involved with a displacement based system.


How about a four wheeled bike like the 4by or the Yamaha prototype bike?
:no:

[CookMySock]

Exactly what I thought, however theres a third option, use a voice-coil actuator and position encoderYes that would be your 'force' rather than displacement servo, but that operates undamped, and now your software has an enormous task ahead of it figuring out what the fuck just happened after the littlest surface irregularity. It may be that you could use a variable steering damper and help it a little, but I think you will find the bars will just fling everywhere out of control.

The difference between force and displacement on a gyro is simple maths, so if you used a stepper motor with transmission, frame x,y,z sensor, measure bar torque (motor current?), I think you could easily calculate what was going to happen. The system would be naturally damped and the stepper could possibly even permit movement from an external shock to the steering if you chose to.

Interesting!

Steve

[Hoon]

RC bikes use a servo to control their steering so I can't see why that wouldn't work if scaled up. The servo steering horn would need to utilize some sort of spring/damping system to absorb bumps. I would leave the actual bump handling to the main sensor/steering feedback loop.
In fact I'd start off first with a 1:5 radio controlled bike, get my sensor/software/control side of things sorted first then scale up to a full version as testing could become a bit of a concern with an unpiloted motorcycle.

[bogan]

Yes that would be your 'force' rather than displacement servo, but that operates undamped, and now your software has an enormous task ahead of it figuring out what the fuck just happened after the littlest surface irregularity. It may be that you could use a variable steering damper and help it a little, but I think you will find the bars will just fling everywhere out of control.

The difference between force and displacement on a gyro is simple maths, so if you used a stepper motor with transmission, frame x,y,z sensor, measure bar torque (motor current?), I think you could easily calculate what was going to happen. The system would be naturally damped and the stepper could possibly even permit movement from an external shock to the steering if you chose to.

Interesting!

Steve

hmmm, think there is still some damping in a voice coil, as wires moving through magnetic feild, so could possibly control damping using the driver circuitry, will have to look into how much could be provided though. Also as the 'bars' are moved the encoder would pick that up and the force could be adjusted to compensate, or I could add a steering damper!

ah fuck, gtg, will look at ur second point later

[Maki]

Why do you say it must be a pressure and not a displacement. (and I stand corrected on this but )If you are putting pressure on the bars, but not getting any displacement out of it, you are not going to be making a difference to the steering geometry or lean angle, and hence no difference to the steering itself?

"I am a master of physics, I agree with dangerousbastard, no point in repeating what he said."

[bogan]

The difference between force and displacement on a gyro is simple maths, so if you used a stepper motor with transmission, frame x,y,z sensor, measure bar torque (motor current?), I think you could easily calculate what was going to happen. The system would be naturally damped and the stepper could possibly even permit movement from an external shock to the steering if you chose to.

steppers are generally used for non-feedback high precision position control (my guess is thats what the above guy uses), so having a bump upset the step count would remove that benefit though you could do as you say with a standard DC motor to measure current/torque just as well. Ill add that to the list of things to look into more:
Voice coil
DC motor
Pnuematics
is what I've got so far.
Yeh the XYZ sensor (I assume that means accelerometer) would be a plan for sensing lean angles, dunno if you remember the traction feedback device thread from bout a year ago? it went over a lot of that side of it.


RC bikes use a servo to control their steering so I can't see why that wouldn't work if scaled up. The servo steering horn would need to utilize some sort of spring/damping system to absorb bumps. I would leave the actual bump handling to the main sensor/steering feedback loop.
In fact I'd start off first with a 1:5 radio controlled bike, get my sensor/software/control side of things sorted first then scale up to a full version as testing could become a bit of a concern with an unpiloted motorcycle.


The steering on this is accomplished by changing the angle of the forks, not the turning of the forks. Are the newer ones like that?
Exactly. The "steering" action is actually free, for some small angle (about 15 degrees either side). When you steer the bike, the steering leans to the desired side, and then the chassis follows, leaning to the same side. Stability is produced with the spin of the front wheel, which acts as a flywheel, with the metal inside the front tire.

sounds like rc bikes use different steering technology to full size bikes, which doesn't sound at all transferable.

[paturoa]

I think you'd need a (twin contra rotating?) gyro measured system too to determine lean and turning.

Would the damping ratio and pressure change with speed as the gyro effect of the front wheel increases?

I agree with the pressure not displacement, as to initiate the turn "counter steering" pressure is required, but the displacement is positive. Interesting control problem, as after the turn starts the pressure would increase as the bars turn the opposite way to the pressure as the turn commences.

Interesting to think this one through!

[bogan]

I think you'd need a (twin contra rotating?) gyro measured system too to determine lean and turning.

Would the damping ratio and pressure change with speed as the gyro effect of the front wheel increases?

I agree with the pressure not displacement, as to initiate the turn "counter steering" pressure is required, but the displacement is positive. Interesting control problem, as after the turn starts the pressure would increase as the bars turn the opposite way to the pressure as the turn commences.

Interesting to think this one through!

Um, contra-rotating gyro would have no net gyro forces wouldn't it?

Pressures etc would increase with more speed, will have a lot of monitoring equipment on the bike so will be easy enough to factor in that increase, or program it to learn itself.

[gatch]

How about a rail gun instead..

[bogan]

How about a rail gun instead..

nah, flatmate called dibs on that idea<_<

[Ocean1]

Um, contra-rotating gyro would have no net gyro forces wouldn't it?

Don't believe so, google gyro platforms, (the physical frame for a gyroscopic compass).

In fact a gyro platform is likely the best way to establish a baseline for the bike control, some good examples in sports photographic systems... like this: http://www.aerialexposures.com/review.htm

[bogan]

Don't believe so, google gyro platforms, (the physical frame for a gyroscopic compass).

In fact a gyro platform is likely the best way to establish a baseline for the bike control, some good examples in sports photographic systems... like this: http://www.aerialexposures.com/review.htm

I was under the impression contra-rotating was two objects rotating on the same axis but in opposite directions, so the gyroscopic precessions would cancel out.

How would I use a gyro platform (one in you link has 3 gyros in it i'm guessing?) to establish a baseline? Do you mean use it as a gyro based tilt sensor?

[Ocean1]

I was under the impression contra-rotating was two objects rotating on the same axis but in opposite directions, so the gyroscopic precessions would cancel out.

It's all mass and it's all rotating, gyro platforms usually have two or more flywheels. Military platforms for weapons guidance are the top end, they spin up huge revs in order to keep the mass low.


How would I use a gyro platform (one in you link has 3 gyros in it i'm guessing?) to establish a baseline? Do you mean use it as a gyro based tilt sensor?

Yes, connect X and Y axis transducers to the platform and feed the outputs to a PLC. The PLC controls the steering based on whatever parameters experimentation shows works. The programme could get a bit complex, and fairly counter-intuitive…

[grusomhat]

Interesting thread. Keep the tech talk coming!

[bogan]

It's all mass and it's all rotating, gyro platforms usually have two or more flywheels. Military platforms for weapons guidance are the top end, they spin up huge revs in order to keep the mass low. yeh but I'm pretty sure they will never be contra-rotating, in fact flywheel energy storage systems used in cars use contra-rotating discs so the gyroscopic effects are canceled out. Its also why motogp bikes (i think they do this now, or it could have just been an idea) have the crank turning the opposite way as the wheels.




Yes, connect X and Y axis transducers to the platform and feed the outputs to a PLC. The PLC controls the steering based on whatever parameters experimentation shows works. The programme could get a bit complex, and fairly counter-intuitive…

Yeh I won't be using a PLC, computers are so cheap and they run C# (FTW), but was considering using a gyro tilt sensor, would get away with a single disc one as bikes heading is irrelevant at this stage. Could get massive precision out of it using air bearings for the frame and 4096 position magnetic encoders.

[CookMySock]

steppers are generally used for non-feedback high precision position control (my guess is thats what the above guy uses), so having a bump upset the step count would remove that benefit though you could do as you say with a standard DC motor to measure current/torque just as well. Ill add that to the list of things to look into more:
Voice coil
DC motor
Pnuematics
is what I've got so far.
Yeh the XYZ sensor (I assume that means accelerometer) would be a plan for sensing lean angles, dunno if you remember the traction feedback device thread from bout a year ago? it went over a lot of that side of it.I'm a bit pissed, but steppers can do a lot more than position things with some processor behind them. Yes you will have to think in advance what will happen - theres no way you will be able to just fire some step commands at a corner, but I think in the end there is a lot more that can be accomplished.

You will have to hydraulic-damp any pneumatics or other force-based systems or bumps are going to be hell. Maybe at high speed it will be smooth, but at low speed it will flop around like a bitch.

With displacement servos you will need to measure steering system torque and follow the front wheel around with the stepper to set torque to the desired amount. So the steering input will be that torque you measured rather than the steering input you created. Clear as mud? So you don't talk to the stepper directly - the torque feedback loop does, and you just input what steering torque you want and the stepper ramps it up.


Steve

[Grasshopperus]

I'd be keen to help out, obviously we'd need to set this up on a gsxr-1000 and make it do a lap of pukekohe or something.

I think you could avoid using a stepper motor by using a pulley system with to control the steering with a gyro and Ping-type sensors on either side of the body to figure out lean angle. It would work similar to the typical 'balancing bot' project that you see guys doing as basic microcontroller programming projects.

Then you could get hardout with a forward-facing laser scanner to map upcoming contours and such to anticipate bumps and dips or just make sure it doesn't run into anything.

It's not just the alcomohol talking, I'm keen.

[Ocean1]

yeh but I'm pretty sure they will never be contra-rotating, in fact flywheel energy storage systems used in cars use contra-rotating discs so the gyroscopic effects are canceled out. Its also why motogp bikes (i think they do this now, or it could have just been an idea) have the crank turning the opposite way as the wheels.

They do use contra-rotating flywheels. I’d say the aim is to eliminate precession issues (the 90deg deviation to external inputs) but retain the stability function.



Yeh I won't be using a PLC, computers are so cheap and they run C# (FTW), but was considering using a gyro tilt sensor, would get away with a single disc one as bikes heading is irrelevant at this stage. Could get massive precision out of it using air bearings for the frame and 4096 position magnetic encoders.

A gyro is the easiest way to get attitude data, an accelerometer won’t work. Yup, get an old laptop for a donor. In fact you could build a tolerably good gyro from the hard drive…

Be much easier if you don’t mind recycling existing systems. Go see New Age Materials on Gracefield Rd and rummage through their stuff. Or Casa Modular Systems in Petone.

[Ocean1]

Oh, and I've got a pneumatic actuator you can have with a 3-15psi control input that would handle steering, just need a wee bottle of air...

[varminter]

Perhaps I've missed the point here, but wouldn't it be easier to just ride it yourself.

[bogan]

I'm a bit pissed, but steppers can do a lot more than position things with some processor behind them. Yes you will have to think in advance what will happen - theres no way you will be able to just fire some step commands at a corner, but I think in the end there is a lot more that can be accomplished.

You will have to hydraulic-damp any pneumatics or other force-based systems or bumps are going to be hell. Maybe at high speed it will be smooth, but at low speed it will flop around like a bitch.

With displacement servos you will need to measure steering system torque and follow the front wheel around with the stepper to set torque to the desired amount. So the steering input will be that torque you measured rather than the steering input you created. Clear as mud? So you don't talk to the stepper directly - the torque feedback loop does, and you just input what steering torque you want and the stepper ramps it up.


Steve

Ok, will add steppers to the shortlist.

I'll see what the damping is like in the chosen steering system before adding another one, the voice-coil may have enough control to do it, will ask my flatmate as he knows more about that sort of thing than I.

Hmmm, not overly sure what you mean with using both...

[bogan]

I'd be keen to help out, obviously we'd need to set this up on a gsxr-1000 and make it do a lap of pukekohe or something.

I think you could avoid using a stepper motor by using a pulley system with to control the steering with a gyro and Ping-type sensors on either side of the body to figure out lean angle. It would work similar to the typical 'balancing bot' project that you see guys doing as basic microcontroller programming projects.

Then you could get hardout with a forward-facing laser scanner to map upcoming contours and such to anticipate bumps and dips or just make sure it doesn't run into anything.

It's not just the alcomohol talking, I'm keen.

yip, that'd be the final goal of the project, dunno if anyone would give me a gixxer atm though. Hmmm, interesting idea with the mechanical control, but I'd want more direct computer control over things than that, the sensors and equipment i plan on using would be able to simulate such a system anyway if needed. Lidar is sooo last decade, machine vision is the way. But yeh, thanks for the offer and chances are I'll take you up on it once the project gets started.

[bogan]

They do use contra-rotating flywheels. I’d say the aim is to eliminate precession issues (the 90deg deviation to external inputs) but retain the stability function.




A gyro is the easiest way to get attitude data, an accelerometer won’t work. Yup, get an old laptop for a donor. In fact you could build a tolerably good gyro from the hard drive…

Be much easier if you don’t mind recycling existing systems. Go see New Age Materials on Gracefield Rd and rummage through their stuff. Or Casa Modular Systems in Petone.

hmmm, I though the two were linked, back to physics school for me!

An accelerometer will work (test rig bout a year ago), but is susceptible to bumps and fast orientation changes, I'm thinking a gyro is the way to go this time.

I have access to a CNC mill so probably will make a really accurate usb gyro module, as it will be the main measurement device. Will salvage the bits for throttle, clutch, brakes, shifter, fold up trainer wheels, etc where possible, but wont skimp on quality for the important stuff, as I've learnt you will only have to redo it later anyway.

Will keep the pneumatic actuator off in mind, thanks.

[bogan]

Perhaps I've missed the point here, but wouldn't it be easier to just ride it yourself.

yup, missed it by a mile ;)

[bogan]

Ok, I'm pretty sure I was right about the contra-rotating discs see bottom of this page (http://www.mariner.connectfree.co.uk/html/gyro.htm) perhaps you are refering to the triple gyro set up (one for each axis)?

[Ocean1]

Those flywheels aren't rotating on the same axis, they're both on an independantly mounted fulcrum.

[Redstar]

effort was made into the creation of a gyroscopic regenerative motor for Train propoltion it worked ,but required a dead straight track as when the gyro was effected laterally in a off line direction the reaction gyro set the train in the oposite direction and that created an wobble that grew into a catatropic situation, its critical to have a compensation drive stabiliser that consists of a conical belt componation drive stabiliser two horizontally opposed conical drives with a soft rubber belt as it gyros it will automatically create a reverse bias that will as least maintain the bike in a straight direction. when the bike is turned there will need to be a fluid gravity pump that sends weight from the down side to the up side the fluid is pumped hydracaccally using the turn solenoids it needs to be a high pressure systems using DOT 3 oil and a high pressure capacity pump. retractable stabalisation minor tractor wheels may be advisable? owing to the relative delay in solenoid activation a pre primed pressure release secondary solenoid is advisable. Gravity plays a big part in stabilisation so keeping the weight low and having the pump that conical inverter lower in the frame is a must. Usually Eddie currents and electromagnetic induction in the electrial system will need a sink earth capacitor say 200 micro farads accross all solonoids or other inductive loads is always a good idea. The unwanted flux measured in Henry.s should not exceed 1.3 Joules thats a watt / per second for the uninitiated otherwise your bike is sound well done.

[bogan]

Those flywheels aren't rotating on the same axis, they're both on an independantly mounted fulcrum.

yeh, they ain't contra-rotating in the pic either, its once of the sentences lower down

Many other arrangements are possible - two contra-rotating flywheels close to each other on the same axis will do the trick; the gyroscopic moments cancel out but the angular moments add

edit: fuck it, 2moro i'm gonna tape some angle grinders together and see what the go is, If i don't post again, it all went horribly wrong :laugh:

[Ocean1]

Bolt propellers to 'em.

And for fuck's sake don't turn them over, you'll dissappear up yer own woirmhole.

[marty]

can't be fucked reading all the above, but at low speed steering is all good, but even my C90 scooter steers via countersteering above about 30km/h.

[bogan]

Bolt propellers to 'em.

And for fuck's sake don't turn them over, you'll dissappear up yer own woirmhole.

That doesn't sound pleasant at all! So ended up bolting them together (using their handles and a bit of plate), really wanna find something that needs grinding like that now. Anyway, no noticeable difference between having none on or both on for moving them round, one on is harder as well as precession being noticeable; so guess they do cancel out.