stair/step climbing capability?

hi, I'm looking at design options for the mobile base of a robot to be used to roam in a house. I'm leaning towards the basic two wheel differential steer with caster(s) as a good design for flat surfaces. However, I`m starting to lose nights of sleep trying to think of a way of providing this base with a stair climbing ability or at least a way to negotiatie a single step, as often found separating rooms in houses. Just cant think of a simple way, other than perhaps using tracks but then that really tears up the floor.

anybody seen elegant and practical solutions for this.....or is it just something to avoid until it becomes practical for the hobbyist to build bipeds ?

thanks in advance for your input.

Reply to
rene.cote
Loading thread data ...

Weren't the daleks fitted with hovercraft type jet sort of things that let them float up stairs in one episode/series? Bit like the harrier.

Whats your budget?

Reply to
Englander

Check out Urbie:

formatting link

- dan michaels

formatting link
=============================

Reply to
dan michaels

There is the tank-tread idea. However, to avoid tearing up the floor, you can use a tread that is all rubber.

biped for the hobbyist will be difficult at best, however the idea of a quadriped might be a workable solution.

if the stairs in question have a solid wall on each side, why not install a track in the walls and allow the robot to connect itself to the track and ride up that track under its own power. (positioning might be a bit problematic).

---Keith Lehman

Reply to
Keith Lehman

This might end up being too complicated than its worth, but let me try to describe it anyway.

I saw a science fiction movie once, where each of a vehicle's normal four wheels were replaced with a grouping of three wheels, each on the end of an arm. The three arms jutted out radially from the center (where the normal drive shaft would be). It appeared that all three wheels for each wheel 'assembly' were driven via a common system (belt, etc), but that the assembly of all three wheels could be independently rotated as well.

If the arms were long enough to put each successive wheel up a stair, this might do the trick, although synchonizing everything might be a real challenge. It would be a great way to negotiate other obstacles as well, though.

- Dave

Reply to
David Gutow

Google for 'Tri-Star' to find more (US Patent 3348518)

Not really designed for this sort of job - more to provide drive in mud/water.

Dave

Reply to
Dave Garnett

I've thought about using treads several times. I haven't had luck finding ready-made treads/drivetrains. I'm sure they're out there.

I've considered building my own using chains. It would be fairly easy to use modified chain plates that are U shaped so that something (like a patch of rubber) could be affixed to outside of the bottom of the U (extending away from the chain.

...or it could provide power to charge the robot...

--kyler

Reply to
Kyler Laird

Oh yeah and of course everyone knows (or should know... the origin of the daleks) they were driven by (the "brains" were) GM KaLeds (previously a humanoid race), OK, so much for science fiction stories I see you think, how often has scifi come true? Progress = vision plus technlogical skill/ or magic as it might have been called in the past and other stuff still is called today.

beware, dont let it happen here, please!

Reply to
Englander

Here's some:

formatting link

-Brian

Reply to
Brian Dean

I found an R/C 1/6th scale M5 Stuart Tank at Walmart. This is $150 and is worth it for the treads and suspension alone. This is a fairly decently sized tread.

You can find pictures at my website.

-- D. Jay Newman

formatting link

Reply to
D. Jay Newman

I suggested something like last time the issue of stair-climbing came around. Someone else said use a pulley-n-hoist system. I suggested using a 1x10" or 1x12" shelving-type board on the steps along the wall

- as a ramp. Put it under the hand-rail so it doesn't intrude so far across the stairs. The bot of course would have to be narrow enuf and strong enuf to negogiate it.

Also, check out Gordon McComb's new Canopus ...

formatting link

- dan michaels

formatting link
=========================

Reply to
dan michaels

Check out the Shrimp:

formatting link
And the Solero:
formatting link

Reply to
Tee

Thanks to all that replied to this query. Lots of good ideas to tackle this problem.

I eliminated the solutions which require a modified environment ( ramps, pulleys, tracks...elevator) which do solve the immediate problem but is not an universal approach and greatly limits portability (from one house to the other).

A think the simplest hobbyist solution right now is probably something like Gord Mccombs "Canopus" tracked platform or a slight variation of it as used in the Urbie ( at least one more sensor and actuator).

The ultimate is probably to reproduce a "Shrimp" like platform but at six DC motors, two servos and a complex structure it is not for the faint at heart. It would be good for outdoors too. The only major issue apart from cost(number of actuartors and control circuits) is the footprint (lenght around 24") which doesnt make it for a nimble platform to negotiate the clearances of a standard home.

As all the pros say: fix an objective for your platform and stick to it.

By the way, the only fail proof option was proposed by my own wife who knows nothing of robotics. It involves buying a new larger home with a single story and no steps/stairs..........

Reply to
rene.cote

Well, actually even a one floor house will have plenty of obstacles that the robot needs to get over. For instance the typical door list at the floor can be hard to get over without tracks, and you still risk that the robot tips over if it is not equipped with any counter-movements.

I think to have a more stable climbing ability it needs to shift its center of gravity to counter for move much like a human do when they lean forward.

I have too thought about how it would be to make a bipedal robot but considering the effort put into it by huge research institutes, and they can hardly make it, I guess it was a bit over the top. Although one thing I did realise was that it had to have more joints than originally thought of, the ability to lean to shift weight. I think a lot of physics is needed to make it move with some grace and retain balance. And I also think that it requires some unknown parameters that has to be learned by the system depending on the constructions weight (Neural nets). Balance -> Move -> Unbalance -> Counter-Move ->

Balance - learning to walk like a baby.

So my next thought was to create a much simpler approach with a robot base that has wheels as its main way of getting around but when encountering stairs it would move forward some "legs" (out of its belly) for support on the first step, before shifting its entire body onto those "legs" and finally retract the wheels up and in again on the next step. The legs could also be pretty soft (beanbags) so that it could get sufficient support on e.g. door lists. I dont think it would be a particularly fast approach but maybe it could work?

Good luck! JC

Reply to
Jeceel

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.