Accelerometers and Attitude Sensors

Look at this.
Mercury switch - Wikipedia, the free encyclopedia

Sorry, I should have pointed out that tilt switches will not work in this application (centrifugal force and all that). Peter Harris

My other thought was some type of pendulum/inclinometer perhaps, but I suppose that would be subject to the same forces.

This is a head scratcher.

Keep in mind that a motorcycle "countersteers". When you are at speed and turn to the left, the handlebar and front tire turns slightly right, hence "countersteer". the same is true for a turn to the right, the handlebar and the front tire turns slightly left. If the headlight is attached to the steering head it will turn away from the direction you are traveling. This is exaserbated by the angle that the light attains. A frame mounted headlight is better but far from ideal. Just a thought! Bob Falk (bobco)

Countersteering only initiates the turn. The turn is maintained by the front wheel pointed in the *turning* direction. It is therefore better to mount the light on the steering head, as it then points the illumination towards the turn. However, the necessary banking of the motorcycle puts the inside turn illumination too close to the ground and the outside turn illumination towards the sky, which is why I want to switch extra lighting on in the turn. Peter Harris

I agree that the useful lighting is lost on the near surface. But disagree that the countersteering doesn't continue through the corner. The countersteering only requires a minute input on the opposite hand grip to initiate; but the harder you push the tighter the turn.

All I can say to that is try riding round in circles (at whatever speed you like) while noting which way the handlebars are pointing. Regards, Peter Harris

I am not an expert rider, but Kieth Code is. Please follow the path and read the article in its entirety. Path: http://www.vf750fd.com/blurbs/countercode.html Also try Googling Countersteer if you question Keith.

"I am not an expert rider" - you are quite correct. "but Kieth Code is" - correct again, apart from the spelling of Keith. "Please follow the path and read the article in its entirety." - if you yourself would only do as you suggest, you would find that the discussion is on whether body-shifting or counter-steering is the more effective method of *initiating* a turn. It says nothing about *maintaining* a turn. Please exert a degree of common sense (what do you expect the slip angle on the front tyre to be?) In any case this has nothing to do with my original query, which is whether anyone knows of a transducer/sensor which is cheaper than an aircraft's artificial horizon but allows lights to be switched on when the bike assumes a leaning attitude. I would be extremely grateful for some information. Peter Harris <1 advanced riding and 2 advanced driving certificates>

I think you're right, It initiates a turn. If it weren't for counter steering, you'd fall to the ground. I wrote a while back on the art of counter steering in a bicycle, if that sheds any light to this confusion ? (same concept, two wheeled vehicle single track) http://cozybeehive.blogspot.com/2008/05/art-of-countersteering.html Also, I'm not a motorcycle rider but what's the nature of the problem exactly? I thought your initial description of it was a little vague? How is the degree of illumination affected by the leaning? Please add some detail, sorry. :) Ron

Hello there, Peter. Welcome to CR4.

I assume this is a road bike, as opposed to dirt bike? If yes, does the countersteer issue really come into play?

My mind drifts towards something like an 'artificial horizon' from aircraft. In lieu of the three dimensions of flight, the issue of roll is the only interest here?

I tried something like this (quite a long time ago), but abandoned the project... I could not resolve the issue of my extraneous light shining into the oncoming traffic. There was somewhat a solution in a handlebar control to disable the lights, but with that in mind, I simply used the control to enable the auxiliary lights. Perhaps an opto-isolator relay, to detect oncoming headlights?

Just thinking out loud... may be of use to you.

Hella® has a driving light that has an internal reflector that is initiated during a turn to illuminate the direction turned, although its for a four wheeled vehicle. It may use some sort of accelerometer, I'm not sure.

Still sticking with the idea of a road bike...

The motorcycle turn is more about leaning the machine than the turn of a piece of control equipment. Turns will result in an apparent centripetal down force, rendering an accelerometer inaccurate or ineffective... I think.

This is a mind bender, to be sure.

What about a Yaw Rate Gyro ?

e.g. http://www.analog.com/en/sensors/inertial-sensors/adxrs150/products/product.html

Regards Woody

Thanks, but the PDF declares "The output signal, RATEOUT (1B, 2A), is a voltage proportional to the angular rate about the axis normal to the top surface of the package". So the signal is proportional to rate of change of position, not to absolute position. Peter Harris

My bad ...

How about http://www.xbow.com/Products/productdetails.aspx?sid=238
or one of their other products?

Cheers Woody

Thanks very much for trying, but that data sheet has "Principle of Operation The CXTA Series Tilt Sensors use a micro-machined acceleration sensing element with a DC response to measure inclination relative to gravity. The response of the tilt sensor depends on the magnitude of gravity parallel to the sensor element. The output of the tilt sensor will be an offset voltage plus the voltage response proportional to the amount of gravity measured by the sensor" so we are back to tilt switches affected by centrifugal force. I don't see any answer except a spinning mechanical gyroscope with electrical contacts, but i was hoping for something cheap. Peter Harris

Not sure what the problem is. Single track vehicles allow you to lean so doesn't the beam stay parallel to the road surface. I can see how this would be a problem with a car.

Here's an idea. Is it possible to implement fiber optics and a motor, and perhaps you could control the servo or stepper through a toggle switch for swiveling motion, depending on which way you corner?

Ofcourse, this will be manual control.

Think of a headlight that is "low beam / dipped" i.e. the top of the beam is cut off so the oncoming traffic is not blinded - when you corner / lean / bank on a motor cycle the horizontal pattern becomes diagonal and minimal light is emmited at the "top" of the beam - this is where the bike is heading into in a corner, giving the rider a "dark spot". The bottom of the beam is also a dark spot as well, preventing early identification of pot holes etc.

One suggestion that others have successfully used is Aux riding lights set up to cross in front of the bike - that way when the bike is leant over, the aux lights provide additional illumination - or else use high beam

I think I see what the OP's problem is. This is a lag problem. The headlight, even if mounted on the front wheel or on the handle bars remains pointed in the direction of the longitudinal axis of the body of the vehicle. So he light beam or the headlight does not anticipate the direction in which the vehicle is turning. I would ask the original poster to check out "Patentstorm" and see some of the ideas out there on adaptive headlamps for two wheelers. Some of those patents have gone past their expiry dates :)

What about using a magnetic flux sensor coupled to a plc that can establish a vertical orientation and then read turn rate and apply servos to rotate the light into the turn as it maintains the earth related oval beam. Getting the wording right is a challenge, thank goodness I wont be concocting the algorithm to drive it. I thought about gyro's, but they become complex with the amount of irregular gyroscopic precession you will have to deal with. The earth's magnetic flux while varying around the globe, is stable enough in regional time frames the plc will be recalculating in. The odd overhead or buried high tension cable you cross could complicate matters for the programmer.

I think you need a 3D accelerometer evaluation kit like this one

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=876-1002-ND

Digikey have got loads. Go to digikey: search for "eval kits" then click on: Eval Boards - Sensors (419 items); then in the "sensor type" column select "accelerometer, 3 axis" (Note**); then "apply filter"; you should be down to about 50 items you can either "view page", or narrow your search.

Note** You really need "Accelerometer, Gyroscope, 3 axis", but the eval kits are a bit pricey (the actual components may be reasonably priced if you're thinking of developing a commercial product).

You might also try to adapt a Wii motion plus remote.

It is only an idea: the integral of a angular rate is an angle.

If you use a rate sensor positioned so that it will be sensitive to the inclination rate of the cycle and integrate its signal the result will be the angle made by the cycle with the vertical. If I rightly understood this is what you need.

You can get such rate sensors for about $10..50 able to measure up to 500°/s.

The output signal is DC low voltage ( 1.25 ± 1.25 V) and you can integrate it with a simple circuit.

I used same approach for other angular measurements where accelerometers were not the solution and obtained errors ≤ 1.5° for angles of 360° obtained in 12x30° steps.

I fear this brings me back to where I started: " The only attitude sensors I can find are in fact accelerometers. Do I have to integrate the acceleration signal or is there an alternative?" Thanks, anyway, to you and to all who have struggled with this recalcitrant problem. Peter Harris

It depends what you define as reference for the attitude. If you use an acceleration field then an accelerometer, as you very well mentioned, gives you the position with respect to the field vector. For you this indication is not what you need since the vector is not anymore vertical but deviated by centrifugal acceleration. The solution I suggested measures angles with respect to the rotation axis considered as reference for the rate sensor. The rate sensor (solid state is based on a Coriolis effect) is TOTALLY insensitive to acceleration. If you place it so that it senses the rotation rate with respect to the horizontal length axis then you get via integration the angle the cycle gets with respect to vertical. Accelerometers are used as attitude sensors usually in quasi static applications in for instance body position in analysis of human behavior.

The rate sensors I think off are mems so what redfred writes and what I suggested are same sensors. As mentioned I used them and you can get even less expensive ones since they are used for camera image stabilization.

How about just having the light itself held on an axle which goes through (well points through) its centre of gravity. It could be very lightly damped and sprung to return it to the horizontal "on average" but when the bike tilts its own inertia holds it steady relative to the horizontal.

Better still, have the light on an axle which points through a point lower than the centre of gravity. Lightly damped and sprung just strongly enough to keep the light at the correct angle.

This will certainly take a variety of iterations to get this right. The natural problem of doing something new.

I would consider a MEMS based gyroscope combined with some single chip embedded controller that turns On/Off some power MOSFET. Unlike a mercury tilt switch, the gyroscope will respond to actual changes in orientation, not just an apparent change in force orientation. The controller will have to sort out these signals and decide when to turn on your auxiliary lights, which light and when to turn them off. I would not try to adjust the angle of your main headlight. You want this system to supplement the main lighting, as you wisely propose.

An initial problem you will have to solve will be changing your zero, vertical reference position (no added light condition) from a just started motorcycle that is resting on a kick stand, to the orientation of moving down the road. I think that this will be a trivial problem to solve with an embedded controller, but this scenario shows why an embedded system will be required.

Hi Peter,

You are correct that countersteering only initiates the turn. It has two effects. The obvious one is the the front wheel must be moved out from under the bike's CG to start a turn. By happy coincidence, the torque caused by gyro procession of the front wheel (when countersteered) also aids in banking. The Motorcycle Safety Council come up with the phrase "push right go right" to highlight the need for countersteering -- which many people are 1. completely unaware of when riding normally, but 2. in a panic situation do just the opposite of what is required, and find the bike very hard to get around the corner.

I am no longer an expert rider, but I was many years ago, and consciously countersteering to bank the bike can get you through a corner faster than the other guy, who thinks he "just leans" into the corner.

I think you could use a small gyro sensor. Typical gravity based tilt sensors etc will always think that "down" is in line with the bike's frame. In endurance racing, we had the headlight, conventionally mounted, and two cornering lights, mounted and aimed for full bank (45 degrees or so).

Another possibility is to simply measure g force vertically (with reference to the frame). You'd need to filter out bumps and high speed hill cresting (etc) but any prolonged vertical g force (with reference to the motorcycle frame) means that you are cornering, and the magnitude of that g force defines the bank angle. Sense steering head angle at a high rate to keep track of the turn direction: a slight steering head rotation to the left, followed by an increase in g force means you are in a turn to the right.

The obvious best g meter: an IPhone... there's an app for that.

Your idea is an excellent one. Our three headlight system worked OK for roadracing, where we spent most of our time fully cranked over, but even then, was not great: in transitions, you were sometimes driving blind -- and there were a couple high speed sweepers where you couldn't go fast enough to require a full bank. Your system could enable active and continuous steering and leveling of the aux lights.

I once owned a Citroen SM, which, in the European version, had steering and leveling headlights. The US version had none of that, sadly, because we had laws requiring the headlights be permanently and rigidly mounted, a holdover from the transition from horse and buggy days, when you could unhook a headlamp, and carry it with you into the house.

Great project. Good luck.

what about using some reversing sensors mounted on the side of the bike

Great suggestion, it seems to me!

That is an absolutely brilliant suggestion, which I will investigate forthwith. Peter Harris

how about using a WIDER and brighter headlight?

and, instead of turning lights up front, use smaller NON-BLINKING lights, because you want a steady illumination up front, and the blinking turning lights only for the vehicles following you.

Sorry, this does not work. Quite apart from the issue of rebuilding the fairing of an expensive standard bike, we have to consider the feelings of other road users. A dipped beam should have a flat top, with or without an extension upwards to read signs on the near side of the road. Any extension upwards on the off side dazzles oncoming traffic. Now when the bike is tilted into a turn, the flat top of the beam makes an angle with the ground. On the near side the illuminated length of road shortens. On the offside the beam points into the air and dazzles. A wider headlight simply magnifies this effect and a brighter headlight is illegal. If you believe that turning lights at the front are unnecessary, then I'm afraid you are mistaken, as oncoming vehicles also need to know when you are turning across them. Peter Harris

the motorcycle was invented in 1885 when the roads were real bad and dark streets were more the norm than the rule. how is it, AFAIK, that up to this time, the issue was never thought of by motorcyclists? I would think that it's because motorcyclists know how to take a curve safely, no matter what time of the day it is.

I admit I'm not a motorcyclist, but I've been a racing cyclist, and in my native place, we have lots of not-well-lighted roads. at night, when turning curves, all I do is SLOW DOWN, so there is no need for me to bank to make a turn. I do not use a headlight on my racer bike, nor do I use a headlamp strapped to my forehead.

another thing I've noticed, when driving a car at night, and you're being blinded by a rude driver who doesn't dip his light, it helps to TURN OFF YOUR headlight. surprisingly, it helps because your eyes will simply use the illumination from the opposite car, and you won't have to struggle to look at the illumination from YOUR headlights.

just my thoughts.

What all this seemingly endless and senseless craving for automation? God has given us the natural senses of sight and balance. Why do we need equipment to do this for us? If one is fit to ride a motorcycle, one should be fit enough to see that he needs more light up front at the proper time, and less light as necessary. Peter, why don't you just install more lights in front that you can switch on as needed? <GROAN>

This post reminds me of this post.

Thank you for that. Yes, I did have a car once with manual advance and retard of the ignition, but it did not run very well. As for "God has given us the natural senses of sight and balance", I will agree that He has, but He left me short of sufficient fingers to operate the switches for extra light as I relax the throttle, apply the brakes and dip the headlight as I enter the corner. For those advanced riders who point out that I should be doing all that in a sequence on the straight, I would agree with them but for the fact that switching on the supplementary illumination before the curve will dazzle oncoming drivers, and doing it in the curve is a distraction. I am not sure how non-riders appreciate that motorcycle controls are not particularly ergonomic as standard, and adding another control to be used in turns is not a very attractive idea. Peter Harris

Gentlemen, Why not put a sensor in the riders helmet and a sensor on the motorcycle connected via bluetooth. A proficient rider keeps his head level with the road. (If you doubt this; watch a roadrace). The difference in the readings would give you an indication of angle that the motorcycle has attained during the direction change. This information could be then used to alter the lighting and respect the oncoming motorist. A speed sensor could be used to prevent the system from actuating at slow speeds. Just a thought from a guy that pushes harder on the opposite handgrip to increase the smoothness of my turns!

Gentlemen, Why not put a sensor in the riders helmet and a sensor on the motorcycle connected via bluetooth.

For several reasons: (1) Why use two sensors when one will do? (2) Why add unnecessary weight to the helmet? (3) Why risk your insurance by fitting an unapproved extension to your safety gear? (4) Why disable the system with yet another sensor? If you have managed to crank the bike over at a low speed in the dark you really do need some help with seeing things. If you have a sensor which would work in this context why not just fit it to the frame? Peter Harris

Peter, it looks as though you have all the answers except one. How to step on another ones toes without ruining their shine. It is called tact.

It's your problem, not mine. Every rider on this forum but you knows that countersteering is only to start a turn, yet you persist. It is not even relevant to my original query. Now you come up with some strange ideas on helmet sensors without even specifying the type of sensor you would use, which sounds more like dreaming than a serious engineering discussion, and again does not answer my original query, which is what type of sensor I can possibly use. I have received some excellent advice from others, for which I am truly grateful, but for the future I suggest you keep to the point and you keep to the facts.

As for "God has given us the natural senses of sight and balance",

One could add that he did not give us the ability to see in the dark. Our race bikes had about three times the wattage of standard road bikes but still left us occasionally having to try to see in very dim light, because of the discrepancy between headlight aim and bank angle. A really good lighting system would have shaved perhaps 1 second per lap off our nighttime lap times, I'd guess. Road riders routinely overdrive their headlights, so a system like you have described would almost certainly save lives. Active headlight aim is again being introduced in cars (and is now legal in the US, unlike in the 1970's, when Citroens here had to be delivered with crummy fixed headlights), but the need is greater for motorcycles, where, when cranked over into a turn, the headlight is only aimed in the vaguest way toward way you need to see.

Yes indeed we have no business going faster than a brisk jog, and if you must use a motorcycle at night I consider it unsporting to use additional illumination. Moonlight should be sufficient for your needs.
Del _{(Teasing mode off)}

God gave us adequate daytime vision, until 40 for me. Litton gave us 4th Generation night vision (and no, you don't get blinded by oncoming lights; that's just a movie myth or the Taliban could just throw road flares at us!). Just don't have a wreck with them, the night vision goggles can handle it but our face doesn't fare well when the goggles get pushed into your helmet! Note: binocular, zero magnification, 4th generation not Walmart grade! Just a thought, rather just a no budget limit ($14,000) thought!

You're turning and probably slowing down some, and you might not be able see very far around the bend, anyway. Is it really that much of an issue if the headlight beam is a bit lower? (I never rode really fast bikes, so I don't know the answer to this. Just asking.)

I think it is a serious issue for people who ride fairly fast. The beam pattern is quite flat, so when the bike is banked, much of the beam is aimed way up in the trees, and another portion is aimed down into the ground too close to the front of the bike. Because even the high beam drops slightly with distance, when banked to the right, the beam aims to the left.

I mentioned that having adaptive lighting would have shaved a second or so off of our lap times, but that was comparing Peter's idea to our race lighting system which used cornering lamps that were on all the time. They worked OK in a full bank and going straight ahead, but not in transitions, and not in turns where we were not cranked all the way over. Had we tried to ride with just the stock light, we would have had to slow down by perhaps 10 seconds per lap (and even then would be riding more dangerously than we would be in the daytime).

However, our system was usable only in a race. On the road we could not use such a system, because the cornering lights would dazzle oncoming traffic.

Even in cars, people routinely drive too fast for their headlights, maintaining high speed even when the beams are dipped (at which point they are adequate for perhaps 45 mph). With bikes being 30 some times as dangerous as cars (statistically, in terms of their ability to kill you) bikes need all the help they can get.

"it is a serious issue for people who ride fairly fast"

IMHO, for such people, who cannot control the urge for speed, they are a danger to themselves and others. what, is slowing down at a curve too feminine?

I don't think OP is talking about racing, too, but he's using it on city streets.

maybe he loves that rush of wind up his skirt?

IMHO, for such people, who cannot control the urge for speed, they are a danger to themselves and others. what, is slowing down at a curve too feminine?

All motorcyclists are a danger to themselves and others -- that goes without saying: last I checked the figure was 38 times more likely to be killed on a bike than in a car (in the US). Some women like to ride fast too -- not because it makes them feel more feminine, but because its fun.

If Peter is concerned mainly about city streets, then I don't personally see the need for additional lighting. In most cities you can get around without any vehicle lights at all. The only time you are substantially over-driving your lights is on twisty roads at speeds beyond those used in cities.

Another thought:

In ordinary car driving, there are many times (especially on two lane roads) where I am driving on blind faith alone. When I have oncoming traffic, I want my beams to be brighter, not dimmer, because my pupils have already closed down slightly because of the oncoming car. But instead, I look down and to the right to avoid the oncoming car affecting my eyesight too much. All I can do is hope there is nothing in the middle of the road up ahead. Even car headlights are not particularly good, and bike headlights are worse.

Dear Peter Harris,

I looked over the thread and determined you really ought to just get a Cadillac and stop riding motorcycles.

I was very impressed at how sweet and kind others of the forum have been towards you.

Thank you for the suggestion, but I will stick to the BMW bike. I agree with you that I have been given some great engineering advice, which I fully intend to take up. The rest I can leave behind. Peter Harris