Does a riderless bike prove anything?
Mon, April 18, 2011 
Engineers over at Cornell University have designed and built a riderless bike which seems to have no other purpose than to debunk the theory that the gyroscopic precession of a spinning wheel, and caster action of the front wheel has nothing to do with the self-steering properties of a bicycle.
I am sorry but this is not a bicycle yet, it is a push toy; when you put pedals and a rider on it, then it will be a bicycle.
I believe that giroscopics and caster action contribute to balance and steering; however, that is not the whole story. Simple momentum has a lot to do with it; take a surf board for example that has no wheels.
A surfer has to initially move his body constantly to remain upright, but when he gets up to speed it becomes easier to maintain balance and his body is practically stationary. A bicycle too, the faster you go the easier it becomes to balance and hold a straight line.
Another natural and intuitive thing to do is to lean into a corner; in the direction we are turning. We even do it while running, and for that matter so do all animals.
While running, if you lean to the left you automatically step to the left to maintain balance, and so you turn; it is an intuitive movement without thought. So if a bicycle is designed to steer in the direction it is leaned, it too becomes intuitive.
It is the intuitive nature of the way a bicycle handles that makes it relatively easy to ride. Riding at slow speed it is almost impossible to ride in a straight line; as the rider falls to the left he steers to the left bringing the bicycle back under his body to maintain balance.
Much in the same way you balance a broom on your hand; you constantly move the hand to keep it under the center of mass to maintain balance.
At very low speeds the cyclist will physically steer the bike left or right by turning the handlebars; as speed increases the rider will steer by leaning to the left or right. This is where the self steering properties of the bicycle kick in.
As the bike falls to the left the head tube of the frame moves to the left; so too does the steering axis.
The steering axis reaches the ground at a point ahead of where the front wheel contacts the ground.
This is known as “Trail,” because the wheel trails along behind the steering axis.
Leaning to the left will cause the front wheel to steer to the left; leaning to the right will steer to the right.
On the Cornell model the exact opposite is true. They deliberately placed the steering axis behind the front wheel’s point of contact. (Negative trail.) They did this in order to prove a point, but in doing so created a reason their model works. As it falls to the left, it steers to the right thus correcting its direction of travel. (Watch the video on their site.)
In order for the Cornell model to work they have added weights. The momentum of a weight on a pole ahead of the contraption pulls it along. Another weight placed low on the front steering falls at a quicker rate causing the front wheel to turn in the opposite direction. (Go to article link above and click on the picture to see an enlargement.)
As I see it all that has been proven here is that by adding carefully placed weights you can make something self-steer even though its steering axis is behind the front wheel’s point of contact.
Gyroscopic precession and caster action have been removed, but this had to be replaced with a system of weights to achieve the same end. It is simply a different way to skin a cat; it proves nothing
As a model this apparatus is a cleaver idea, but will not work as a bicycle; here is why. As I have already mentioned to balance on a bicycle as you fall to the left you instinctively steer to the left to bring the bicycle back under the center of mass, which is the rider.
On the Cornell model as you fall to the left it steers to the right, the rider will be unceremoniously dumped in the road. Even worse as you intuitively lean into a corner the bike will steer in the opposite direction.
Not only have Cornell taken away the intuitive handling nature of the bicycle, they have designed an unridable bicycle. Don’t believe me? Build it and see.
There is some more reading on Head Angles and Steering here, and on Trail, fork rake and a little bit of history here.
Reader Comments (16)
I know from my motorcycle roadracing days that countersteering is the way to change from travelling in a straight line to leaned over in a curve. Before and after the steering input, whether straight up and going straight, or leaned over and following a constant radius arc, you can take your hands off the bars. Making a fast transition means you really have to countersteer forcefully.
At speed, a bicycle is similar, but the gyroscopic force is smaller. Maybe I have to try taking my hands off the bars and leaning over a bit to be sure, but I think I know how my roadbike will react to this -- continue travelling straight.
I agree that the Cornell device will behave completely differently, one won't know how it behaves untill someone is sitting on it and trying to steer it.
From reading the paper, it doesn't appear they eliminated caster from their model to necessary disprove the caster effect, but merely to show that other dynamics besides the caster effect can be used .
I fail to understand what the real purpose of this exercise is; It is like looking for a problem where there is no problem. Surely there are plenty of real problems in the world that need solving. Who cares how a bicycle performs without its rider? Bicycles are never called on to go anywhere without a rider.
Dave
To me it doesn't look like its achieving anything other than a waste of money and time.
Sam
Hello Dave and everyone – I’m excited that a forum like this is discussing our work, because of your extensive bicycle design experience!
I want to explain what we set out to show. If I succeed, hopefully you will find the results as interesting as we did. I will also react to some things you say which, if I understand them, I at least partly disagree with. Apologies for the length, but I was concerned that if I answered more tersely, it might lead to misunderstandings.
By the way, to see the 4 page paper, the 50 page appendix, and 40 page historical study, along with videos and links to other videos, please go to http://ruina.tam.cornell.edu/research/topics/bicycle_mechanics/stablebicycle/ or http://bicycle.tudelft.nl/stablebicycle/ , supposedly identical sites.
A word about my background: I had some fun racing (never very well) in the early 70’s, then went to England hoping to learn frame-building. I didn’t get along all that well with Harry Quinn; and Johnny Berry died just as something was about to start. After that I fell into a path of looking for bicycle answers or insights through the engineering sciences. That has been a long hard road, trying to figure out the physics, and the human factors (psychology), what is important and what is possible. I’m still looking. Fortunately Andy Ruina at Cornell, Arend Schwab and Jodi Kooijman at Delft, and Jaap Meijaard now at Twente University share this interest. (As does the bike group at UC Davis.)
OUR RESEARCH
This crazy experimental bike was not intended as a model for future bikes! And we were very careful to state that even something demonstrably self-stable might not be considered easily rideable or even stable by a human rider. Mainly we were reacting to a variety of common statements: (1) that negative trail would be difficult or impossible to ride because the wheel would unstably try to flip around; (2) that the steering would not automatically turn into a fall; and (3) that a gyro is essential for bicycle self-balance. Are those common statements, found in many books on bicycles and motorcycles, true?
We are reporting that if taken at face value, they do not appear to be accurate. That doesn’t mean we advocate a change in bike design, but we’d like to see the understanding and logic improved. And, possibly it will turn out that designers need not be afraid of zero trail or slightly negative trail, or may gain some benefit from thinking about mass placement when luggage or rider position is at issue. To paraphrase the motorcycle engineer Wilson-Jones in the 1950’s, he played with negative trail and found it was far less problematic than everyone had predicted.
The foregoing is a general explanation of our motivation: to test the reasoning which argues that trail is valuable. (We think that reasoning is at least incomplete, or maybe even wrong.) Other motivations are to interpret everyday ideas into precise technical language, to test them effectively, and to develop software that genuinely predicts bicycle self-corrections.
Maybe it will turn out that we had the wrong idea of what people REALLY mean when they talk about the consequences of front end geometry! But our way to find that out is to do as we did, and then put the results out for feedback.
RESPONSES
Now I will try to address some specific statements:
“Does a riderless bike prove anything?”
We think it proves that some common ideas [which present negative trail on a two-wheel vehicle as making stability impossible] are wrong, or let’s say incomplete. Our bike had negative trail, yet the front wheel did not flip around, and the bicycle was still able to steer to the right when it fell to the right.
“seems to have no other purpose than to debunk the theory that the gyroscopic precession of a spinning wheel, and caster action of the front wheel has NOTHING to do [typo? We meant to debunk the theory that caster and gyro actions are NECESSARY for bicycle stability] with the self-steering properties of a bicycle.”
As I have re-interpreted those words, yes we were trying to understand and debunk that theory (as well as other things, such as proving that our software can make valid predictions). I think we succeeded. We proved that without trail or gyro contribution you can still have stability! We would never have attempted this if bicycle and motorcycle design philosophy stated “negative trail MIGHT tend to create instability UNLESS the mass distribution is appropriate”, etc.. But we have only ever seen unqualified stronger statements, e.g. that stability is impossible if trail is negative.
“this is not a bicycle yet”. True. And it will probably never be a bicycle, unless someone gets curious about its riding properties! It might feel and behave terribly. Its only purpose was to examine the clearest statements and arguments about stability that we have heard.
“I believe that giroscopics and caster action contribute to balance and steering; however, that is not the whole story.”
That’s for sure! We don’t yet know what a rider likes or most easily adapts to. We were only focusing in on the narrower statement that negative trail (say) causes instability. Now, we also might wonder if negative trail is so bad to ride by a person, but we never tested it.
“Simple momentum has a lot to do with it; take a surf board for example that has no wheels.”
I don’t fully accept the statement. For sure, if you have a fast moving bike with the steering welded, it will fall over in the same short time whether moving or standing still. Our way of understanding is that falling over should initiate turning, which can restore balance if you are moving fast enough. Skateboards can be stable this way, when moving fast enough. (If you put a rigid mannequin on a surfboard, is it stable? We never looked at that.)
“It is the intuitive nature of the way a bicycle handles that makes it relatively easy to ride.”
I bet you are right. But that is not what we were looking at. If you want to see the problem from our eyes, look through everything that is written specifically about the relation of trail and / or a spinning wheel to stability. What does it say, and what does it mean? I think you could significantly help our understanding by performing that interpretation. The way in which a bicycle responds to human intentions may have little or nothing to do with the stability we were studying! By the way, some research that especially interests me is that by Tony Doyle, a former RAF squadron leader and aerobatic instructor. We have his thesis and article online. Here is a little note of what he attempted: http://books.google.com/books?id=dG6WPVidI-wC page 36 or search on Doyle. Here is a paper: http://audiophile.tam.cornell.edu/~als93/Bicycle/doyle1988essential.pdf
“Riding at slow speed it is almost impossible to ride in a straight line; as the rider falls to the left he steers to the left bringing the bicycle back under his body to maintain balance.”
Sure. And with a stable riderless bike, the same is true. You should look at the gyro-bike: with extra gyro effect, it does that task at low speed without a rider. (And it is not the first to make that point.) http://www.thegyrobike.com/Gyrobike-HOW-IT-WORKS-s/83.htm
“At very low speeds the cyclist will physically steer the bike left or right by turning the handlebars; as speed increases the rider will steer by leaning to the left or right. This is where the self steering properties of the bicycle kick in.”
I ask that you give a more precise explanation here. Do you mean the rider steers by leaning her upper body relative to the bike frame? Or that the whole system, leaning over, then steers itself to catch the fall? Please clarify.
“Leaning to the left will cause the front wheel to steer to the left; leaning to the right will steer to the right.”
Again, what do you mean? If you are riding straight, leaning your body left relative to the bike will make the frame lean to the right. And for most bikes, they will then steer to the right. But in contrast, if a bike FALLS left (let’s say while at rest, so it is easy to observe), whether the steering turns left (desirable), right (undesirable) or not at all is based on a bunch of factors. Trail does not have any unique effect here the way that Jones and others thought. We can have positive trail causing reversed steering in a fall.
“As it [TMS bike] falls to the left, it steers to the right thus correcting its direction of travel.”
Is that a typo? The TMS bike steers to the left to catch a leftward fall.
“The momentum of a weight on a pole ahead of the contraption pulls it along.”
I don’t really agree with that technically, even though I sometimes fall into that style of thinking. A bike on a treadmill, at rest in the laboratory, has exactly the same steering dynamics. See the riderless bike treadmill experiments here: http://bicycle.tudelft.nl/schwab/Bicycle/index.htm
“Another weight placed low on the front steering falls at a quicker rate causing the front wheel to turn in the opposite direction.”
Don’t you mean, in the SAME direction as the fall, if it is ahead of the steer axis…? Please explain.
“As I see it all that has been proven here is that by adding carefully placed weights you can make something self-steer even though its steering axis is behind the front wheel’s point of contact.”
Pretty much, that is true! So trail is not absolutely essential for stability, even though it might be important for other reasons. And there are still other things one might do also to achieve self-steering. Remember, all we are trying to do is point out that negative trail, despite what has been written about it, is not NECESSARILY catastrophic, and indeed the self-steering aspects were never properly understood.
“As a model this apparatus is a cleaver idea, but will not work as a bicycle; “… “On the Cornell model as you fall to the left it steers to the right”
I disagree, it works because when falling right it steers right. Whether you have a real rider or just a lead mass, it should balance itself the same.
“Even worse as you intuitively lean into a corner the bike will steer in the opposite direction.”
Please define ‘lean into a corner’. Do you mean, lean your torso relative to the frame, as we do in no-hands riding? Or do you mean lean the bike over for a corner? We agree with old observations that trail helps cause the steering to turn left if you bend your torso rightward relative to the frame (which tilts the frame left). That looks like part of the reason we can control a bike with torso motions. But the front assembly mass (wheel and handlebars) also play an important role, so if the fork has a forward offset, you still get this same effect even if trail is zero. (The no-hands control factor is shown by riding a straight line while leaning your upper body to the right of the frame. The steering should try to turn left, and you should be able to feel it.)
Reader Comment: “From reading the paper, it doesn't appear they eliminated caster from their model to necessary disprove the caster effect, but merely to show that other dynamics besides the caster effect can be used .”
It was a mix of aims. We definitely hoped to show that other factors could play an important role. But we were also taking aim at the statement that a negative caster was inevitably unstable. And we discovered near the end, that positive trail doesn’t necessarily make the steering turn the way we’d like in a fall.
“I fail to understand what the real purpose of this exercise is; It is like looking for a problem where there is no problem. Surely there are plenty of real problems in the world that need solving. Who cares how a bicycle performs without its rider? Bicycles are never called on to go anywhere without a rider.”
I think that’s a little unfair. We are looking for understanding. And judging by the number of ‘explanations’ of the importance of trail, some others are interested in it too. We want to address the body of work that attempts to explain bicycle design, test it, and update it where it doesn’t fit the facts.
I also have some feeling or hope that this work will eventually lead to more-stable recumbent, better bikes for old and wobbly riders, and maybe also more maneuverable racing bikes. And hopefully, in the bigger picture we can better relate human behavior to the tools that will best help us in our lives. But even if none of those benefits materialize, I am hoping to achieve real understanding and quantitative tools, so explanations and predictions can be accurate.
CONCLUDING
Thanks for an opportunity to address the kinds of questions that people have about our result! There are many other things in the paper, including internet links to historical documents, and other counterexamples such as a stable rear-steer design. So we hope people take a look and come back with their thoughts.
Sincerely yours,
Jim Papadopoulos
Green Bay
Jim, Thats EASY for you to say? I think!
Jim,
Thank you for your long and detailed explanation of what you are doing. It will take me some time to answer it all point by point, but here are the observations I made.
In the video I see the apparatus being taken through two separate test runs. In the first I see it fall (Lean) to the right and turn to the right. Then it steers to the left and corrects the fall; it pulls itself upright again. In the second test run it falls and turns to the left, then corrects itself by turning to the right.
Did I observe correctly, or did I miss something? In each case when I said a rider leans into a corner, I meant the rider leans both his/her body and the bike. And if a bike performed as your model did, a lean to the left would mean the bike would steer to the right. (Based on what I saw in the video.)
Before I go any further, I want to ask my readers if they really want to know. Because one thing surprised and kind of disappointed me a little about this post was the lack of comments. Whereas my previous post, a simple story about a bike ride, brought 22 comments.
Jim, if you and the people at Cornell are really interested in my opinion and input, I would be happy to do that. A phone conversation would probably achieve more than going through a lengthy post here and boring the pants off my readers, who for the most part like me just want to ride their bikes.
Dave
Dave, yes, I definitely want to kow, this is interesting stuff.
Jim, while you have apparently proved that caster and gyroscopic effect are not inherently necessary for stability in the riderless two-wheeled vehicle in your study, you have not proven their necessity in bicycles as we know them today, riderless or otherwise. Your vehicle bears no similarity to a modern safety bicycle; therefore, it cannot prove anything about how they work. Interesting science? Yes, certainly, but not appropo to bicycle design as we know it. Could it possibly lead to something useful for HPV's? Sure, but I still want to know how my bicycle really works.
>Jim, while you have apparently proved that caster and
>gyroscopic effect are not inherently necessary for stability
>in the riderless two-wheeled vehicle in your study, you
>have not proven their necessity in bicycles as we know
>them today, riderless or otherwise. Your vehicle bears no
>similarity to a modern safety bicycle; therefore, it cannot
>prove anything about how they work. Interesting science?
>Yes, certainly, but not appropo to bicycle design as we
>know it. Could it possibly lead to something useful for
>HPV's? Sure, but I still want to know how my bicycle really works.
Point taken. Our analyses (in the online appendix) show that for a fairly typical bike, canceling either the trail or the gyro contribution eliminates self-stability.
But I don't personally look at this as the end of the story. How good is a bike that is not quite self-stable? What is it we really like in a bike, and is there a way to get 'more' of it without traditional compromises? Asking the basic questions is one way to get an understanding of the whole ball of wax. This published study is just one step along the way.
To put it another way, I think we can build on this and other work, by examining actual human needs or preferences, and possibly gain a much better understanding of how regular bikes work so well.
Jim Papadopoulos
Dear Dave, you wrote:
>Did I observe correctly, or did I miss something?
>In each case when I said a rider leans into a corner,
>I meant the rider leans both his/her body and the bike.
>And if a bike performed as your model did, a lean to
>the left would mean the bike would steer to the right.
>(Based on what I saw in the video.)
I will have to get back to the video, try to see what you saw, and reply. I will do that on email or phone because you are right, long detail-filled scribblings will not interest most readers!
Unfortunately this won't be right away, because my mind is too filled with other things. But I don't want to drop it.
Sincerely yours,
Jim Papadopoulos
"Disappointed in lack of comments": Understand but for most (?) of us, we are simply interested in cycling safely and what we have now is performing beautifully.
The reason for the study sounds like a challenge of assumptions. Which is what science does. If no one challenges assumptions then, in this case, bike designers are locked into a box of set parameters. The actual application of the findings may not cahnge design based on human factors but we may beable to design a better bike or style of bike knowing that previous base assumptions don't necessarily need to be followed as rigid dogma.
Ralph:
Andy here, a co-author of the Science Bike paper.
In short, what you write above is exactly our purpose, hope and goal.
If you want to see pics, videos, and read an infinite amount about this
funny TMS bicycle and related bicycle ideas, go to my www page:
http://ruina.tam.cornell.edu/research/topics/bicycle_mechanics/stablebicycle/
This, in turn, links to Arend Schwabb's pages which have even more
information.
not long ago, i dreamt of riderless tandem... weird.
Marvelous discussion. I am very happy that Jim P. has shown up to respond to reader comments. I totally disagree with posters who question why research like this ought to be conducted; or worse, whine that our bikes are fine as is, as if no one should be asking questions about them. The fact is, there is a lot that we don't understand about bikes and bicycle steering. The bicycle has evolved through a little knowledge and a lot of tinkering, and although it is a superb machine there are still certainly improvements to be made to both the way they handle, and to our understanding of *why* they handle that way. This kind of careful analysis is a great way to find some small parts of the solution to both these problems.
Guys, nice discussion showing the difference between basic science and applied engineering. Sort of like philosophy and law.
Anyway, for me the study explains why you have to sit upright for no hands riding. Letting go of the handlebar whilst staying hunched forward in the steering position results in an unstable ride, sitting up straight to put more weight on the back wheel and heighten the centre of mass paradoxically results in a stable ride, and this research resolves the paradox.
Basic science often has no obvious practical application until engineers and business people make things from on the basics. Electricity has been known and useless for for millennia. Strangely, when it comes to rewards, the business type end up rich, engineers middle class, and scientists begging for funding.