h a l f b a k e r yPoof of concept
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
I still recall sitting on my fathers lap to steer the car on a country lane and the surprise at the seeming lack of logic that after turning the wheel to execute a turn, one must then turn it back to the original position. Though several weeks have passed since then, I still wonder if steering wouldnt
be easier with a point and drive design.
Thus turning the steering wheel would turn the front wheels to the same degree, but after the car has executed the turn that same angle determined by a gyro compass, the wheels straighten out but not the steering wheel. After turning a 90 degree right corner, the front of the steering wheel would end up at the three oclock position.
One long even curve would demand continuous small steering wheel adjustments with the new system instead of holding it at a fixed poaition, but a curvy or straight road would need half as many steering movements and no need to release the wheel.
[link]
|
|
Can we not go the whole hog and have auto-drive akin to jet plane controls? A knob which sets the direction of travel according to the degrees of a compass.
Another croissant for [FJ]. I have suspisions you collect all these croissants and sell them at your local WI market for a nice profit. |
|
|
...and by looking at the N, E, S and
W letters marked around the edge
of the steering wheel you'd always
know what direction you were
travelling in. |
|
|
Sorry, have to fishbone this. ?ow
is the car supposed to know when
to stop turning? You'd need
something to watch the road, and
some heavy computation, and then
you're into DARPA territory. |
|
|
The car would stop turning when it was pointing at the spot equivalent to where the wheel was turned. Imagine clock numbers on your steering wheel, 12 at the top. So you turn the wheel so that 2 is at the top, meaning you want the car to head in the 2 o'clock direction. The car turns until it's pointing to 2 o'clock and stops turning. |
|
|
Motorcycles. They are fairly self-correcting when it comes to steering. That is, unless of course you REALLY turn it, then it just wants to fall over. But is its like a lane change or simple turns, then its all point-and-shoot. |
|
|
/Though several weeks have passed since then/ |
|
|
How would you arrange for the indicators to cancel? |
|
|
I thought this was a new Reality TV show: What You Steal Is What You Get |
|
|
Actually, it could also be a new high concept steakhouse |
|
|
[egbert] I would give about the same answer as waugs; when the car knows the turn is completed, the wheels are allowed to return to neutral and the turn indicators would cancel. |
|
|
As far as laughter is concerned, this is a site where one receives much more joy than can possibly be repaid. |
|
|
I realized, while thinking about my clock-numbers example, that one would actually turn the wheel in the opposite direction than one does now. Going from straight (12 o'clock) to 2 o'clock is a slight turn to the right in practice, however you'd need to turn the wheel counter (anti) clockwise, which is sort of left-ish, to bring the 2 to the top. |
|
|
That's a feature, not a bug. |
|
|
How is one to convey the RATE at
which the vehicle turns? - for that
is what a steering wheel is, a
control of the rate of turn. |
|
|
To provide the rate, one would
have to make a turn by gently
turning the wheel all throughout
the curve, ever careful not to allow
the wheels to begin to return to
the straight ahead. Very hard, very
dangerous. |
|
|
//RATE// I dont see any difference from conventional steering where the wheels also immediately respond to a rotation of the steering wheel and where the correct vehicle speed is important. In this case, one would have to continue turning the steering wheel (2 X 45 degrees, 3 X 30 degrees, etc.) to negotiate a standard corner or let up now and then for a gentle curve (longer radius). |
|
|
Today's steering also demands constant adjustment to stay in the lane on a curve. Admittedly the idea seems difficult, based to some degree on learned mindset. |
|
|
// To provide the rate, one would have to make a turn by gently turning the wheel all throughout the curve, ever careful not to allow the wheels to begin to return to the straight ahead. Very hard, very dangerous. // |
|
|
Not at all. You turn the wheel gradually as you go through the turn. 12:15, 12:30, etc... until your wheel (and the car) are pointing at 2. How this is more dangerous than current steering methods is not evident to me. |
|
|
The only difference is you'd need to keep turning the wheel as you negotiate the curve. An argument could be made that this actually gives you *more* control. |
|
|
Fishbone - I wouldn't like the lack of feedback... It would feel like a viscous coupling to the wheels, not my cup o' tea.... |
|
|
// how does the car know you're *completely* done turning? // |
|
|
It stops turning when you stop turning the wheel. Folks are seeing some difficulty in this that I'm not picking up on. |
|
|
I am with [half] on this one. The problem as I see it, [waugs], is that I can sense the action of my steering IMMEDIATELY on the road in good conditions. With this scenario I am obliged to wait until the steering direction that I have 'dialled in' (lets face it, we don't need a steering wheel under this idea's conditions) has decided for itself that the dialled in turn has been completed. The 'feel' of the turn has been turned over to a machine based on the best guess by a human to the point where you have to guess where the machine will stop. It appears to add a superfluous layer. A car with no power steering will always give you the best indication of the road. A car with power steering is designed to slowly move the wheel in the direction and is soon mastered. This steering system is designed to make it easier for small adjustments on straight roads but will hugely compicate judgement based corners. Maybe this is why I can't play video driving games. |
|
|
Imagine a horizontal steering
wheel. The "North" position on the
steering wheel keeps pointing
North, no matter what - you're
effectively turning the car around
the unchangingly oriented steering
wheel. |
|
|
Ah. Now I see where you fellows are off. You are thinking that the car responds in some willy-nilly fashion to the wheel turn, responding in its own time, eventually pointing in the direction you turn it to. That's not how I see it. |
|
|
If you turn the wheel immediately from 12 to 2 in one abrupt jerk of the wheel, the car would immediately reorient itself, steering the wheels to be pointing to 2 as quickly as possible, very likely turning you off the road in the process. Not what you want in a gradual curve situation. You'd 'dial' the car through the turn with slow moves as you progressed. The car will still be going where you point it to go, and I would think "wheel-feel' would remain as responsive as it is now. Possibly even more so. |
|
|
There's no guesswork as to when the car stops turning. It stops when it is going in the direction set by the wheel. |
|
|
Exactly that [half] and better explained. Can you go in the sim please because I really am crap at them!. |
|
|
//There's no guesswork as to when the car stops turning. It stops when it is going in the direction set by the wheel.// I think what we are saying is that the guesswork must be made at the start. 'Its a 10:30 corner..click'. Even if I dial this in I must worry about the possible misjudgement. Even if it is an incremental change I must wait until it is finished (which is not under my control). |
|
|
[half + gnome] Good question. Say you start a curve to the right with a 15 degree increment. If one had instead turned the wheel for 20 degrees, one would notice after a moment (the wheels are still angled) that the car was continuing too far to the right and would turn back the wheel a bit to realign. If one had only turned the wheel for 10 degrees, one would notice after a moment (when the wheels straighten out) that the car was wandering to the left and would turn it a little more. The same thing would happen at the end of the increment if one had turned the wheel exactly 15 degrees. |
|
|
All this seems similar to todays steering, except that now one finds an angle (which could be wrong) such as 10 degrees, makes adjustments in the curve and returns to the neutral position afterwards. With WYSIWYG for the same curve, one continuously increases the angle (say up to 30 degrees) with adjustments and the steering wheel stays at 30 afterwards. |
|
|
[hippo] I see the directions as follows: Turn from north to east; the steering wheel ends up pointing south. Turn south; the wheel points north. Turn west; the wheel points south. Turn north; the wheel points north. |
|
|
A pointer on the steering wheel would function as an in-car compass until one skidded. |
|
|
// Even if it is an incremental change I must wait until it is finished (which is not under my control). // |
|
|
Yes it is under your control. There is the misunderstanding. The car goes where you point it. |
|
|
As FJ illustrates, feeling your way through a curve with this sort of steering would be no different than current ways. When you turn the wheel, you can sense where the car is going. You can still correct the steering if it feels under or over... |
|
|
I'm tired of trying to explain this. There is clearly a fundamental misunderstanding that can't be fixed. |
|
|
I think some of the folks are still confused by what is actually being controlled (and how) by the steering wheel in this idea. I'll throw one more explanation which just might help. Or then again, it could just muddy the waters further. |
|
|
For reference, in a traditional steering system the turn radius (how tightly one turns) is determined by the angle of the tires relative to the longitudinal (back to front) axis of the car. That angle at any moment is approximately proportional to the angle of rotation of the steering wheel at that moment. |
|
|
As written (ignoring annotations), the idea gives the impression that while turning, the angle of the tires would be either some constant set at the factory or some variable determined by a computer. That implies the driver would not have control over the sharpness of the turn, except maybe indirectly through the car's forward speed. If that were the intended case, then this idea would be bone worthy because a driver's primary goal is to follow a given path, not to achieve a particular compass bearing. |
|
|
However, there is yet hope for this idea. With the proper changes to the system, the angle of the tires could instead be set proportional to the instantaneous rate of turn of the steering wheel. That means in order to follow a gentle curve, one would rotate the steering wheel very slowly and stop rotating when the road straightens out. A sharp turn would require a much faster rotation of the steering wheel. If the proportionality constant were carefully selected, then the total change of bearing of the car would be approximately equal to the total amount of rotation of the steering wheel. |
|
|
Since this would almost certainly need to be a drive-by-wire system anyway, the function relating tire angle and rate of steering wheel rotation could be as complicated as you like to get a better match between steering wheel position and the car's bearing. |
|
|
Having said all that, I'm convinced that this is still a bad idea for general use. Although arguably more intuitive to the novice, this type of steering would deliver poorer performance. This is because a human can control the position of the steering wheel more easily, precisely, and consistently than a human can control its speed of rotation. |
|
|
I'm getting what you say [BB] and thank you for elucidating. I guess I am in general against the idea because of the perception of lack of directness of control because a lot of vid games use this pricipal and I can't get on with them. Thank you, however, for tolerating my inane ramblings so far! (ta [waug] too!) <-----------Discrete line drawn underneath! |
|
|
You dont need to modify the car to accomplish this idea. You could hold the steering wheel with specially wheeled motorized driving gloves. The system could have an external computerized sensor pack (gyros, accelerometers, etc.), that connects to the gloves. Turn the car, then hold your hands steady, and the gloves rotate the steering wheel back in place, for you. |
|
|
Well, [bigbrother] your explaination
makes it even more apparent that
it is boneworthy! |
|
|
What if you want to sharply turn
while parking slowly but only for a
short while, e.g. in parallel
parking? |
|
|
Even ignoring parking (theer are
systems to do that for you now
anyway), what if you have already
moved your wheel to the desired
direction then realised you wanted
to turn faster or slower? You are
buggered! You cannot unwind
without telling the car to turn
AWAY...for how could it know that
you mean turn slower or actually
turn away (e.g. to avoid
something). |
|
|
I beleive you are trying to put too
much information through a very
limited channel. |
|
|
It is also, IMHO, counter-intuitive,
as brakes, steering and gas/
accellerator all work with the same
model: more input = higher rate of
change, unwind/take off equals
reduce rate of change to zero (of
course there is engine
braking...unless you have an old
Saab...) |
|
|
this is why we all need slot cars |
|
|
this is not what i expected when i saw the word 'steer' in the title and that farmer john had wrote it. |
|
| |