[HN Gopher] The Brennan self-balancing monorail [video]
___________________________________________________________________
The Brennan self-balancing monorail [video]
Author : Fervicus
Score : 134 points
Date : 2024-02-12 01:51 UTC (21 hours ago)
(HTM) web link (www.youtube.com)
(TXT) w3m dump (www.youtube.com)
| Animats wrote:
| There's a model railroader who builds these things.[1] They
| balance and corner fine.
|
| With one rail, the models can't take power from the rails. They
| run on AA batteries.
|
| There's also a drivable 2-wheeled gyro-balanced car from the late
| 1960s.[2]
|
| These things work, but wobble too much. With active control,
| though...
|
| [1]
| https://www.youtube.com/watch?v=ifroNPpK9jQ&list=PLE006F2D73...
|
| [2] https://www.youtube.com/watch?v=TTCVn4EByfI
| jacquesm wrote:
| That second one is just gorgeous. That guy has every reason to
| look as smug as he does!
| JKCalhoun wrote:
| Yeah, not a stretch to imagine someone could make a Star Wars
| Landspeeder version.
| dsego wrote:
| And here I thought that the gyro motorcycle that appeared a
| decade ago was a novel invention. https://youtu.be/jICGl9jmulc
| Levitating wrote:
| Wait, this isn't a Tom Scott video?
| franky47 wrote:
| That's what I was expecting too.
| toxik wrote:
| Didn't he quit?
| Ballas wrote:
| He didn't quit making videos entirely, he only "quit"
| releasing them to a set schedule.
| Fervicus wrote:
| Quite remarkable. Too ahead of its time? Wonder why it never got
| revisited
| modeless wrote:
| Seems like the requirement for gyroscopes in every car would be
| a deal breaker.
| neom wrote:
| Could you make a gyro car? Akin to a locomotive? I don't
| exactly know how it would work but I was thinking we have
| locomotives so we don't need to put engines in every car. The
| thing I'm thinking about is that the gyro is kinda both a
| measurement device in a way, and also the system that
| actuates. Could these systems be separated, simplified, and
| scaled-down? I don't know, I'm not an engineer, just thinking
| aloud. :)
| jerf wrote:
| With the technology of the time I do not believe there is a
| solution.
|
| Modern technology _might_ be able to turn the car couplings
| into something that can actively manage their tilt relative
| to each other, thus allowing them to propagate the balance
| beyond a single gyro car, but there 's a lot of PID-
| controller-type [1] math that would go into determining if
| this was even possible with real, physical objects trying
| to control things the size and mass of train cars in a
| train track environment. The computations are probably not
| that difficult in 2024 but trying to actually deliver the
| correct forces in a timely manner may be difficult if not
| impossible. There aren't always solutions, or practical
| solutions, to the PID equations in the real world.
|
| Plus, I'm sitting here imagining the size of the electric
| motors we're trying to torque the cars against each other
| with and it's hard not to say that we're better off just
| having two tracks and putting those motors on the wheel
| themselves. (Which then collapses to having a locomotive
| setup like we have now.) In the "power/control/price"
| triangle we're basically forced to take the max on power
| and control, so these are going to be ferociously expensive
| if they can be built at all.
|
| And there's some bad failure cases that I just don't know
| that you can mitigate. Having 30 minutes after power loss
| to evacuate passengers is generally going to be enough, but
| cargo can't be evacuated on that timeline. And that's not
| the only failure case. A seized gyro is going to be
| catastrophic and even a single one could take the whole
| train down, even with that PID control system (which is
| going to be a stretch even without also having to build in
| buffers for failed gyros).
|
| The fact that a 2-track-train can just sit there for
| extended periods of time with no power is one of those
| advantages you don't even think about until you try to take
| it away.
|
| Guy was a genius though. It's basically an analog computer
| that keeps itself upright that you can ride. The final
| design is amazing with its judo-esque approach to using
| physics against itself. But I don't think it can be scaled
| up much beyond "neat tourist attraction". I'd ride one as a
| tourist attraction, though.
|
| [1]: https://en.wikipedia.org/wiki/Proportional%E2%80%93int
| egral%...
| neom wrote:
| Can always count on you for a fantastic reply Jeremy.
| Thank you.
|
| Indeed, the more I think on it the more difficult the
| edge cases seem, I hadn't even gotten to the power out
| thing, although from thinking on the spinning cube
| conversations in this thread[1], I had thought about
| magnets as a failsafe mechanism, but at that point....
|
| "The fact that a 2-track-train can just sit there for
| extended periods of time with no power is one of those
| advantages you don't even think about until you try to
| take it away." <- This heh.
|
| [1] https://news.ycombinator.com/item?id=39336139
| meche123 wrote:
| Because you have to replicate the same complicated mechanism
| for all the cars. Moreover, if a gyro in one of cars
| malfunctions the whole train is at risk.
|
| Another problem is that the system is reactive, meaning that
| there is a lot of shaking and balancing that is quite
| unacceptable - think about drinking a cup of tea in such a
| wobbly train even on straight lines, while on curves it's not
| staying horizontal by design.
|
| Yet another problem is that it can't share the current
| infrastructure without major switch changes while sharing the
| infrastructure the other way around is impossible.
|
| And in the end all of this effort for trying to fix a problem
| that doesn't really exist, at least not in Europe where most of
| the rail infrastructure was already built 100 years ago.
| Reason077 wrote:
| > _"while on curves it 's not staying horizontal by design"_
|
| Worth noting that modern tilting trains do this quite
| successfully (on two rails!), in order to run faster on
| tracks with tight curves. By "leaning in" to curves, you make
| the ride _more_ comfortable, not less.
|
| Tilting trains are used extensively on the UK's West Coast
| Main Line, for example.
| danhor wrote:
| > By "leaning in" to curves, you make the ride more
| comfortable, not less.
|
| That is the intention, but many people report discomfort
| due to the non-ideal adjustments, at least for the tilting
| trains in germany.
| philjohn wrote:
| I'm one of those people.
|
| I commute 50 mins each way on a Pendolino in the UK (into
| and out of London) and find that if I'm doing work on my
| laptop, or responding to emails on my phone as the train
| tilts I feel uncomfortable and dizzy.
|
| The "slow" Desiro trains that don't tilt take 2 minutes
| longer to get into London, because whilst their top speed
| is 15Mph lower they accelerate faster.
| waqf wrote:
| Do you still feel uncomfortable if you cannot see outside
| the train at all (i.e. not even in peripheral vision)? My
| understanding was that the effect was caused by the
| scenery outside the window moving in a way that's
| incommensurate with what your sense of balance is telling
| you.
| georgesimon wrote:
| Ideal adjustments (full compensation) are more
| uncomfortable.
|
| Thought to be so, because there's a mismatch been visual
| and inner ear inputs. That is: you can see you're going
| around a corner, but you don't feel it. Resulting in
| motion sickness. Since all people have different
| tolerances of motion sickness, it's impossible to have a
| "correct" amount.
|
| Almost all systems compromise between zero and full
| compensation, but passenger comfort is just one of many
| factors, and not the most significant. The discomfort on
| German tilting trains is probably because they are too
| close to "ideal".
| danhor wrote:
| Ideal (in the sense of fully compensated) is worse, but
| ideal in the sense of at the right time, with smooth
| interior motion, is much better. A "perfect" tilting
| train shouldn't feel much different than superelevation
| to the same level, but it does.
|
| One of the issues with the German tilting trains is that
| they try to compensate up to 8deg without track knowledge
| (BR 612) thus the first carriage (the one after that less
| so) swinging abruptly to compensate when entering or
| exiting a curve. Not sure how the class 390s solve that
| problem.
| georgesimon wrote:
| So you're saying the rate of change of compensation (or
| "jerk rate" when entering and exiting a curve, as people
| enjoy calling it) is worse on tilting trains?
|
| I would love to see some data - I assumed this was a
| solved problem, but I wouldn't be too surprised if the
| tolerances were revised to fit the problem.
|
| I forgot to mention the 'interesting' part of this
| problem: that the effects are measured cumulatively. That
| is, against passenger journey length. Drivers and other
| staff are not considered as far as I know!
| 1letterunixname wrote:
| And there's zero redundancy. If 1 of those piston seals or
| valves fail, the whole train is at risk for catastrophic
| failure.
|
| Clever engineering without a necessary, defensible,
| competitive business advantage usually becomes either a hobby
| or an art project.
| georgesimon wrote:
| And if one rail has a gap or defect.
| georgesimon wrote:
| Yes, the part of the video that said this would work on
| existing infrastructure is entirely wrong.
|
| It could be interesting with it's own infrastructure, though
| my first thought is that it might be better suited to
| tunnels, being able to maximize use of the tunnel profile.
| 1letterunixname wrote:
| No. Watch the video.
| woah wrote:
| What's the benefit?
| joshuaheard wrote:
| In the video it explains it only uses half the track material
| since it only has one track instead of two. It can also go
| faster since it leans into corners and doesn't derail like
| two track trains.
| kbenson wrote:
| It also required the complex gyro for each car, not just
| the engine, which is why the video states ultimately it was
| killed. I'm not sure the situation has changed with that.
| p1mrx wrote:
| > leans into corners
|
| Modern trains do that too.
| https://en.wikipedia.org/wiki/Advanced_Passenger_Train
| eventually led to
| https://en.wikipedia.org/wiki/Avelia_Liberty, which is
| being deployed for Acela in the US.
| kevin_thibedeau wrote:
| The double flange wheel wouldn't work with existing rail
| switches.
| Jolter wrote:
| Well, that was the most annoying commercial break ever.
| rsanek wrote:
| https://github.com/ajayyy/SponsorBlock
| philsnow wrote:
| yt-dlp has sponsorblock integration, just saying
| btbuildem wrote:
| Yeah that was pretty jarring
| tareqak wrote:
| Do e-scooters and the like self-balance in this way?
|
| It would be cool to have self-balancing rollerblades.
| Jolter wrote:
| Anything with rolling wheels will have some tendency to
| continue straight once the wheels are spinning, but they don't
| help to stabilize in corners. On small wheels, the effect is
| very small in any case, especially since the scooter and its
| batteries are relatively quite heavy.
| arijun wrote:
| The gyroscopic effect of wheels has a minimal impact on the
| tendency of things like bicycles to remain balanced; the
| majority of the effect comes from the rake of the front wheel
| (i.e. that the point where it touches the ground is ahead of
| the pivot point of the front post).
|
| That gyroscopic effect of the wheel works well when the whole
| device is wheel (think hoop rolling), but when you start
| adding weight that's not spinning (like someone riding a
| scooter), the effect falls off.
|
| Edit: it seems that neither the gyroscopic effect nor rake
| are sufficient to explain bicycles' self-stability:
| https://en.wikipedia.org/wiki/Two-mass-skate_bicycle
| phyzome wrote:
| << the point where it touches the ground is ahead of the
| pivot point of the front post >>
|
| The other way around -- the contact patch of the tire is
| behind the projection of the head tube onto the ground.
|
| I love that there are actually three different stability
| mechanisms. I wonder if there are others!
| mring33621 wrote:
| see https://2swiftboards.com/, which is an electric
| scooter with no handlebars.
|
| At low speeds, a high-friction headset adds stability by
| damping the tendency to turn and flop.
|
| At higher speeds, the reverse fork rake and spinning
| wheels provide stability.
|
| This is my read on it, as someone that uses a 2Swift as a
| commuter vehicle, but the physics-based explanation may
| differ.
| FrustratedMonky wrote:
| Incredible. Sometimes it seems like we were more advanced in the
| past.
| phyzome wrote:
| Where do you get that from? We still come up with crazy designs
| all the time (that are rejected in favor of robustness,
| compatibility, etc.)
| JKCalhoun wrote:
| The crazy designs from 100+ years ago though were more
| impressive though because of the limitations imposed from not
| yet having electronics, computers, software.... They had to
| be incredibly clever mechanically. I too was impressed from
| the video the way the guy iterated in clever ways to address
| issues that came up. And, as I say, iterated within a
| relatively narrow realm of options.
|
| Ads a counter-example, I watch hobbyists on YouTube these
| days that build and fly various R/C contraptions that would
| not have been do-able in the days of small gasoline engines
| and huge, heavy radio receivers. It's like today you can take
| a sheet of foam from Lowes, strap on some high efficiency
| brushless DC motors, double-sticky-tape on an ESC, receiver
| and be flying that sheet of foam by the afternoon. Too easy.
| FrustratedMonky wrote:
| There must be enough people that were similarly amazed and/or
| fascinated to vote it to front page. Maybe my phrasing was
| not to your liking.
|
| I'm sorry you've lost your sense of wonder.
| dudeinjapan wrote:
| It reminds me a bit of TRIZ thinking
| https://en.wikipedia.org/wiki/TRIZ. It's not that we were "more
| advanced", it's just that Brennan had a clear-headed approach
| to use mix of elements (gyroscopes, differential/asymmetry,
| pneumatics)
| FrustratedMonky wrote:
| yeah, these old mechanical control devices seem like a real
| life 'steam punk'.
| samatman wrote:
| This is occasionally true, but that is very infrequent.
|
| The example I like to cite is swords. Despite swords playing no
| role in modern combat, modern swords (the real ones, not
| knockoff/costume swords) are of a higher quality, fit, and
| finish, than anything a knight or samurai would have been able
| to obtain. While being dramatically cheaper, in the sense that
| someone would have many fewer hours to work to obtain one.
|
| There are counterexamples of products which would have been
| cheaper at a given quality in the past, or readily available
| when they're difficult to obtain in the present. But once you
| rule out things made with ivory and whalebone, these are scarce
| indeed.
| kabouseng wrote:
| I wonder if gyro drift would eventually cause problems?
| croes wrote:
| What happens on a power failure?
| ethagknight wrote:
| The video says the gyros would continue spinning for 30 minutes
| (inertia) in the event of power failure. Enough time to get the
| kickstand down I guess.
| ethagknight wrote:
| Some things im curious about with this concept.
|
| 1) why not mount the gyro horizontally? 2) would the addition of
| a very large (say 10ft diameter) primary load bearing wheels
| (like a bicycle) assist with stability at speed 3) video mentions
| a need for gyros in each car, I wonder if such a system could
| only require a complicated active control system in the front
| (and maybe rear) car, and the rest could have simpler 'passive'
| gyro arrangements to assist in relative stability to the active
| controls? 4) could modern control systems materially improve upon
| this? For example, an all-electric drive for the intermediate
| gyros off a primary power unit is pretty trivial. 5) why wouldn't
| they want the flywheels higher up, like mounted over the roof?
| quicker reaction time and more force per rotation?
|
| Thats all for now.
___________________________________________________________________
(page generated 2024-02-12 23:02 UTC)