work in progress - hex walker

Very sweet, how much do you expect it to weigh? Do you have any walking algorithms?

I've always avoided walking robots because it seems like they have a higher power requirement than a rolling one. With wheels one only needs to move the robot, with legs, you have to lift it as well.

A few kilos, not sure exactly yet. It will depend on the batteries. I can shed some weight by cutting away some of the base if I need to (drill large holes to make a honeycombe base). Or I could swap the servos for more powerful ones. I just wanted to stop planning and start building :-)

No not yet. I have a few in mind. I'll suspend it off the ground on a tether and experiment by sending it lists of position / time settings, then get it to play the lists.

Yes but I was looking for a control application that emphasised real time control of many motors. I plan to make the step by step build details available once I get it walking. I'll also make the executable available so you can program a PIC yourself without needing an expensive controller.

Regards Sergio Masci

- optimising PIC compiler FREE for personal non-commercial use

If money is not much of an object, the NiMH batteries have a very high power density. It is about 6 lbs (5 lbs 13 ounces) for a 12V 7AH lead acid battery. For NiMH, you can get AA batteries that have 2300 mAh (2.3 AH) they weigh about one ounce each. You will need 3 banks of 10 batteries to be 12.5 volts at 6.9 AH and is about 2 lbs (1 lbs 14 ounces), less than a third of the weight.

$60 dollars for the NiMH (assuming $2.00 per battery) $14 dollars for the lead acid battery.

So, for about 4X the money, you can get 1/3 the weight.

Thank you for the info. I have been thinking about NiMH. I have seen 1800mAH for half the price of 2200mAH (from the same manufacturer). I have also been thinking about Li-Ion batteries. There seem to be quite a few sources of "cheap" Li-Ion for mobile phones on eBay. The biggest hurdle here would be a charger.

Actually I wouldn't need 12.5v, 7.2v would be fine - even less weight :-)

Regards Sergio Masci

- optimising PIC compiler FREE for personal non-commercial use

batteries.

.................

............

I have 3 walkers, and all use NiMH AA-cells, 1800-2100 mAh. These work well, are light in weight, and will run the smaller bots for quite a while. Gimlee, Nico, and Nico2 ...

Lead-acid batteries are probably much too heavy for a walker, especially given the energy capacity of this type of battery. I would probably go with NiMH "C-cells" rather than use a 7.2AH SLA.

I see a lot of walkers on the net made of machined aluminum, but this is also very heavy for a walker. Heavy base requires stronger servos, and these in turn require larger batteries. It's a vicious circle. Anymore, the first thing I try to do is design a light-weight frame, meaning some kind of plastic instead of aluminum. Gimlee uses an aluminum plate for body, but it's basically underpowered for its weight.

have fun,

- dan michaels =====================

You only have to lift some of the legs. The rest of the mass does not have to rise.

m

Actually, most hexapod walkers have to expend some energy just holding themselves up, and will slump down when you turn off the power. This is one of the prices you pay for having cantilevered legs. Eg, get down on the floor and extend your arms & legs out away from your body, and try holding yourself up with only your fingertips and toes touching the floor. One of the advantages of vertebrate skeleton design [except for some amphibians and reptiles] is that vertebrate legs are "rotated" under the body so the bones+skeleton directly hold more of the weight and the muscles can do less work.

Did you look at some of R. Brooks work at MIT AI Lab in the 90s. They did some stuff on behaviu=oural robotics. I think the legged robot was called Hannibal and it used a 'subsumption' architecture.

Your robot looks cool.

Sergio Masci wrote:

Maybe cantilevered legs are not a good design for larger robots just as they are not for larger animals? I suspect a reptile rests on its tummy when not running about? Perhaps the design is only suitable for small robots.

The other problem I see with this design in larger robots is the surface area it takes up.

Also in a domesticated robot a rather large mechanical spider scampering down the passage would be rather scary :)

A domestic robot needs to be "cute" and innocuous looking.

It should also be silent or at least have a nice motor sound.

-John

Sergio, What are those white spacers used on the clear plastic servo brackets? Do they have some kind of flex in them?

-- Gordon

Dan, As I know you've found in your research on legged critters, some insects let their bodies slide on the ground when they walk, and others lower themselves during rest. (Also applies to some quadrupeds, like lizards.) I don't think it would kill a walking robot to have a low-friction glide point on its under carriage, if not during the walking phase at least during the stop phase. A rounded piece of UHMW is rugged, cheap, not that heavy, and has self-lubricating properties.

-- Gordon

I did a nifty Algo for the EH3-R from Lynx Motion. It does inverse kinematics for each leg, with a sort of "coathanger" gait. It does vectored motion, so it can "crab" in any direction. Added a wireless playstation controller. Lots of floating point math, but some good ideas there.

I put Li Ion batteries on it from Lightflightrc.com , along with a voltage comparator circuit to keep from frying them. LiPoly would be better, but the budget didn't allow at the time.

Videos at

I was going to list it on Ebay this weekend, to sponsor the next wave of robotic madness, but now I just don't know...

Mike

Always wanted to do worm drives on a walker for good static power consumption. Of course the trade off is lousy efficiency durring running.

This is exactly the point. When is the last time you saw a 400 kg spider? [heaven forbid]. An 80-tonne dino with cantilevered legs is probably physically impossible, especially when the femurs of those guys were somewhat wider and taller than a man.

BTW, you might be interested that crocs can gallop ... pretty phenomenal to watch ...

You notice that most of the japanese android designs are rather non-threatening looking.

Hi Mike, I notice you don't believe in small AVI files ;-). I downloaded maytag, and it looks pretty much as I expected! Oh, man. [actually, my HD ran out of space this morning and I spent half the day doing file maintainance. Bummer].

At any rate, I was going to mention that Sergio's hexagonal leg attachment design was probably going to involve a lot of trig-math, which you no doubt have been enmeshed in with EH3. Do you think that design is superior to the usual hexapod leg-attachment scheme? It's probably on the same level of verstaility as a synchrodrive, turn on a dime, etc ....

but I imagine you need to solve a lot of maths before it can take even a couple of simple steps. No ????

This would basically work for any geartrain that is not reversible. Most servos don't have that steep of a gear ratio, so they'll backdrive under weight. A rack-and-pinion is not (readily) reversible, is pretty efficient as there is no sliding action (as is the case with a worm gear), and is a natural for converting rotational motion to linear movement. Best of all it's pretty cheap.

A hint of what I've been working on: I have some pull-cord cars I bought at the toy store that have a relatively nice flexible plastic rack gear, about 8" long. The pinion gear it mates with can be fastened to a servo horn, as it has a large hub (with screw holes even!). I cement the rack along the edge of some leg pieces, cut from 1/4" PVC. The legs slide up and down fairly well, and the motors hold a static weight. I've made only one prototype, but I think a full walker could work.

The original intent of the design was for a parallel gripper strong enough to hold a 12 oz. brewski with a standard servo, and not spill it when driving across the living room. (The store-bought cans, not the stuff Mikey (Kap'n Salty) brews in his basement.)

-- Gordon

Hi Gordon, you're obviously correct that real animals with cantilevered legs or side-attachment legs [like lizards, etc] can slide on their bellies, but I've never actually seen any hexapod robots using this idea. They tend to hold their bodies up off the ground, and require energy expenditure to do this. OTOH, 2 or 3 years back I ran across a link to a turtle-like robot that slid on its stomach and waved its feet for locomotion, not unlike swimming. Lost the link, however. So, at least one person caught onto that idea.

Another thing along these lines. The newer version of Urbie has wheels on leg-like extensions it can put down. Doesn't walk, but it's a wheeled/tracked hybrid ...

I've also thought some about the idea of a hybrid vehicle, but with small motorized wheels attached to the end of regular robot legs. On smooth surfaces you roll, and on rough surfaces you walk over them.

Dan, They have some pretty big ones in Australia! Some scorpions, the largest arachnids on the planet, can grow to eight inches long.

Even the small ones tend to be considered the "creepiest things on the planet." I have to agree with John that legged robots are not conducive to acceptance by the average person. When they're holding still they look harmless enough, but when they start to move I can see the goosebumps on people's arms.

We're getting back into black window season around here. If you'd like, I can scoop you up a couple and mail them to you. You can study them all you want!

-- Gordon

Kind of like Shrimp III with the addition of tracks. Build for speed, but can climb over things. I can see that on smooth surfaces of other planets the wheeled motion would mean less battery drain.

Reminds me of the Packbot, but with wheels added to the ends of the outrigger tracks.

-- Gordon