Does anyone know about a robot able climbing stairs that was build as a
I have an about 30cm(10") long board, which needs to be balanced
I have two ideas to acquire this goal:
1. Mount a large piece of metal on it, which can be moved forth and back
by a motor.
2. Have two containers of water, where water can be pumped between them.
Which one is the better?
Any help apreciated,
Water could be a problem if it starts sloshing about, you'll end up with the
thing oscillating like mad, or tipping slightly, then all the water moving
in the direction of tilt and pushing it still further. However, if you can
build tanks so as to avoid this problem and can pump the water fast enough
to compensate for tilting, I'd say they could be better than shifting a huge
weight - for one thing, the weight could only easily be shifted along one
axis, whereas various tanks could be placed anywhere you liked and could
probably shift the centre of mass anywhere you wanteed within the shell.
Also, you'd probably have to sacrifice a HUGE amount of valuable chassis
space inside the machine to make a cavity or tunnel for the weight to move
through, and you'd also need the weight mounted on very strong inflexible
mountings, because any acceleration of the chassis will cause the
acceleration of the large weight to put enormous forces on its mountings,
while water tanks and pipes can be squeezed in around any other internal
systems and can be simply bolted directly to the chassis without messing
about with heavy duty bearings and things.
As for stair climbing, this has always been a tough nut to crack in the
hobbyist budget and skill range. I'm working on a modification to Gordon
McComb's "walkerbot" to allow it to climb stairs (and maybe any other lumpy
surface) easily and also keeping the chassis itself horizontal at all times.
The basic principle is sound and easy to work out, but the implementation is
a nightmare - I need to build a non-symmetrical pantograph mechanism for
each leg in order to raise and lower the feet by about 8" whilst keeping the
pattern they describe when driven by the main motors the same -
unfortunately, I have to choose between two evils: The first is to use a
straight four bar pantograph, with a minimum of awkward joints, but each leg
will stick out by 9" on either side of the machine when fully retracted,
making it far too wide to navigate the average staircase. The second is to
double up the pantograph like lazy tongs, reducing the width of the machine
while retracted, but increasing the number of joints that can work loose,
wobble, or weaken the legs' lateral strength. I still haven't decided what
to do about that one.
I would have gone with the water, but that's a matter of personal style, I
suppose. Everyone tackles a problem in their own way.
It's probably become a standard because of the high popularity of the book.
It's also relatively cheap and easy to build, compared to certain more
conventional mechanisms like caterpillar tracks, which require lots of
bearings, and perfect precision if you don't want the tracks popping off the
rollers all the time. The walker only really needs a lot of angle girders
and a few chains and sprockets (though I found a way to build mine with
nothing but aluminium girders, strips and bolts, avoiding chains which are
almost as tricky to work with as tracks).
Gordon himself frequents this group from time to time, and I don't mean to
cause any offence, but the design, while an excellent starting point and
quite easy to build, does have a couple of flaws, at least in my experience.
Others may have been more succesful than I have. It requires careful
synchronism of the two drive motors to prevent excessive jolting (and I only
made the crank radius on my machine half an inch - I can only imagine how
bad the jerkiness must have been on the original, which will have had even
bigger strides). Built straight from the plans, it is also impossible to
turn the machine around without disturbing this synchronisation, so it can
only easily walk in a straight line without extra modifications.
I added two extra legs, in the centre of the front and back faces of the
machine, moving perpendicularly to the six primary legs, each on an
independent motor. By programming a combination of movements in a
microcontroller to convert simple commands like "left wheel, right turn,
forward, etc" into leg sequences, it can be arranged so that only the two
auxiliary legs and the two central ones on the sides are in contact with the
floor, all acting to produce a turning moment about an axis in the centre of
the machine, allowing it to turn fairly gracefully on the spot. Well,
that's the hypothesis - the mechanical hardware is ready, but I haven't
started on the microcontroller yet. The problem with this system is it
limits the machine to a series of motions not unlike an army drill squad -
it can only turn on the spot through fairly large increments, or paces, in
order to reallign the legs in the right position to then take a forward or
For perfectionists out there, another problem is a certain amount of energy
will be wasted when the machine walks in the vertical motion of the machine
as the legs push up and down - it will also put quite a bit of strain on the
motors as the machine gets heavier when you start adding subsystems.
Perhaps it could be fixed with an arrangement of a carefully machined cam on
each leg adjusting the length so that the motion of the feet while in
contact with the ground is purely horizontal. The vertical motion might
also cause a few problems if you want the machine to serve drinks or
anything like that...
Just a few suggestions, hope they help. As for pictures, I'm afraid I
haven't got any of my machine yet, and probably wont until it's finished.
If I do take any, I'll try to remember to post them here.
It's in both both editions of my book, Robot Builder's Bonanza (first
edition printed 1986). I've seen some pics posted from time to time over
the years by people who have built their own, but most are school-based,
so the sites come and go. It's a VERY large robot -- almost three feet
long. It wouldn't work for stair-stepping as-is, but there are some mods
like using a "tri-star" wheel arrangement that might work, if the wheels
are large enough for the stair risers. Apart from the mechanics of stair
climbing is the horsepower requirements. Stairs are basically 45 degree
inclines to a robot...lots of power needed to get up one of those.
Author: Constructing Robot Bases,
Robot Builder's Sourcebook, Robot Builder's Bonanza
Just a thought .... but maybe one might add a special "helper arm" to
the front of a walker to help boost it up the steps. Basically, would
be a bar held by 2 outriggers [that would rotate around in synch], and
essentially "pull" the rest of the bot up the steps, kinda like a baby
using its front hands/arms to pull its rear end along. Also, something
like the front-end flippers used in Urbie. The bar would only engage
for climbing over things, and might need to be retractable, but the
mechanics would be simpler than jointed legs/etc. Still take some
power, but it could be geared very low.
I whipped up a quick drawing to illustrate - [not a very good draw
program for this] - shows 3 different positions for the helper arm:
- dan michaels
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.