Hacker News new | past | comments | ask | show | jobs | submit login
Walking Truck (wikipedia.org)
134 points by colinmegill 8 days ago | hide | past | web | favorite | 70 comments





> The stepping of the robot was controlled by a human operator through foot and hand movements coupled to hydraulic valves.

> The walking truck was one of the first technological hardware design applications to incorporate force feed-back to give the operator a feel of what was happening.

This might sound primitive in today's autonomy-driven mindset, but in contrast to autonomous robots, this is a technology that could be delivered today (or was already delivered in 1965, according to the article).

So if infantry really does need hardware to move stuff from A to B, the 1965 approach -- primitive as it is -- might be the simplest solution.


> So if infantry really does need hardware to move stuff from A to B, the 1965 approach -- primitive as it is -- might be the simplest solution.

Putting threads on a box or crap on human backs?

Because whether at rest or in action (an other commenter provided a video) that thing looks like it would tip over if you looked at it funny, let alone "over rough terrain".


>Because whether at rest or in action (an other commenter provided a video) that thing looks like it would tip over if you looked at it funny, let alone "over rough terrain".

A group of infantrymen should have no problem getting it back upright. Shenanigans moving heavy things is what they do when they get bored but don't have any lower effort mischief to engage in.


> A group of infantrymen should have no problem getting it back upright.

Well sure but why would they use it in action aside from the fun factor of seeing this lumbering contraption (and at 1400kg empty with a top speed of 8km/h "lumbering" is putting it midly) flail and fall over every few minutes?


They're not gonna walk it out into a firefight but it makes sense for getting heavier weapons into the hard to get to places you inevitably wind up wanting them. This kind of vehicle is useful for moving all the weapons systems we put on top of M113 and Humvee chassis. Also, it's worth remembering that this is a prototype, not a finished system.

> They're not gonna walk it out into a firefight but it makes sense for getting heavier weapons into the hard to get to places you inevitably wind up wanting them.

No better than a souped up ATV would. Probably worse, with harder maintenance and more breakdown potential.

> Also, it's worth remembering that this is a prototype, not a finished system.

Doesn't matter, it's the concept itself which is nonsense.


Go watch some rock crawling videos on YouTube. That's the kind of terrain legged vehicles are envisioned for. Making wheeled vehicles navigate that kind of terrain is non-trivial and requires lots of design compromises. There's a reason you see most people doing it in dedicated recreational vehicles.

>No better than a souped up ATV would

Big obstacles necessitate some combination of big tires (and all the big heavy parts to turn those without breaking) and massive amounts of suspension travel in order to maintain wheels on the ground (and therefore stability). A legged vehicle of the same weight would be able to carry much larger cargo or go over much rougher terrain because it would simply be able to step through each obstacle rather than roll over it.

>Probably worse, with harder maintenance and more breakdown potential.

Pin and bushings joints aren't exactly maintenance intensive and in "you should have replaced this last year" condition they tend to be more operable rotary power transmission systems. The problem is figuring out how to control the damn thing. Fairly recent advances in programming techniques (ML, basically) have resulted in new solutions for this problem which has resulted in renewed interest in legged systems.

>Doesn't matter, it's the concept itself which is nonsense.

There's a reason people keep trying to make legged designs viable. They're very attractive for navigating rough terrain because they offer the potential to do equivalent work with a smaller (and therefore cheaper and simpler to maintain) vehicle. It just turns out that every time someone has tried to build one the technology isn't quite there yet.


> Doesn't matter, it's the concept itself which is nonsense.

Not at all. Look at the kind of terrain that recent Boston Dynamics bots can handle. The concept is solid, they just didn't have the technology yet to make it practical. (And we're still working on it.)


You sound like someone who's never been on a hike, especially on a rock scramble. Try driving any wheeled vehicle over terrain that's filled with rocks and boulders; it can't be done. But I can easily walk over it, climb through fissures, etc.

> You sound like someone who's never been on a hike,

You sound like someone who doesn't understand the concept of context.

> Try driving any wheeled vehicle over terrain that's filled with rocks and boulders; it can't be done.

Neither can driving the Walking Truck over that same terrain is my point.

> But I can easily walk over it, climb through fissures, etc.

Putting crap on the back of humans who are pretty good at navigating bad terrain is exactly one of my suggestions?


Humans can't carry hundreds of pounds of cargo over rocks. A walking truck potentially can (once someone engineers one with a good control system).

>Neither can driving the Walking Truck over that same terrain is my point.

Why do you believe this? You're stating it as fact, and it sounds like BS to me.


> a top speed of 8km/h "lumbering" is putting it midly

8kph is a gentle jog, hardly lumbering.


I suspect you're right - I also suspect modern technology could make such a thing more practical to use - as the original was difficult and tiring to control according to the article.

I'm sure that approach is simpler, but from what I can tell, the (semi?) autonomous technology can also be delivered today: https://www.bostondynamics.com/ls3

I thought of that example, too, but thought that it had been cancelled as unsuccessful.

Turns out it was cancelled for mainly other reasons. Quoting [1]: "By late 2015, the Marines had put the LS3 into storage because of limitations with the robot including loud noise, challenges in repairing it if it breaks, and how to integrate it into a traditional Marine patrol"

[1] https://en.wikipedia.org/wiki/Legged_Squad_Support_System


Turns out the marines don't really need a loud, expensive, fragile machine to haul their gear when they can just use a quad bike.

Robots are awesome but integrating practical robotics with mission critical operations is really really hard, especially because you're automatically competing with battle-tested best in class solutions.


Actually turns out Boston Dynamics had a good idea and over-engineered it. LM built a cheaper, wheeled, semi-autonomous troop support vehicle that ended up selling pretty well.

https://defense-update.com/20070529_smss.html


That thing looks an awful lot like a quad bike with two extra wheels.

AKA a "6 Wheeler". At least that's what we called them when I was kid. They use differential steering, which makes them a good fit for automation. Don't have to worry about turning a steering column, just shift the power around which the off the shelf ATV already handles.

Yep, there were ads for them in the back of Popular Science magazine. They were fully amphibious too and could float.

Don't get me wrong, I love my fellow kinematically-complicated meatbags but I do kind of think wheels win this round.


There are still a couple of companies that sell 6WD (and 8WD) vehicles like these - Argo is one of them (and that machine up-thread looks like they just strapped some stuff to one of those):

https://argoxtv.com/

The other kind of "extreme off-road differential steered" ATV that people both want (and then seemingly want to sell soon after - almost like a boat) is the Russian Sherp:

https://sherpatv.com/

I'd love to have one of those - but they are a bit out of my budget (and my wife would probably divorce me if I brought one home, even if I could afford it).


I love the Sherp! Also the roughly-equivalent bike version: https://canadamotoguide.com/2016/08/03/russian-2wd-bike-is-a...

yeah basically, but with no driver. That was my point about over-engineering: the army didn't want an expensive re-invention of off road vehicles, just existing technology tailored to their needs.

BD has a long term vision for robotics and I'm not suggesting that they should stop working towards that. Just making the point that it's possible to create robust solutions using simpler robotics even in complex environments.


Could you explain what you mean by "selling pretty well"?

It seems they shipped a few prototypes…


The first four block 1 varients deployed to Afghanistan were pretty widely publicized. After that you can find public references to test events throughout the 2010s, the manufacturing of a few block 2 varients, and stated interest from more than one agency. But no there aren't a lot of hard numbers to be found in a Google search. Presumably you could learn more by looking through the actual contracts, but that takes some time and effort.

For a completely new military program, that's close to your best case scenario. Most new programs get dropped after the first prototype.

https://www.auvsi.org/arms-armed-and-armor-lockheed-martin-p...


I saw that Lockheed was not selected by the Army for their next round.

https://www.defensenews.com/land/2017/12/14/four-companies-a...

I guess I don't see delivering prototypes as selling well, but arguing opinions is boring, so whatever.


However, wouldn't the hydraulic + operator version be a) cheaper to develop and b) less complex to build and maintain (useful when deployed) ? I think there is a good case for KISS here.

KISS is slapping an engine and threads or 4~6 all-terrain wheels on a box.

You've made the same comment twice and I only just now managed to figure out that you meant treads, not threads. I thought you maybe meant some kind of box that could be screwed through the soil...


> You've made the same comment twice and I only just now managed to figure out that you meant treads, not threads.

Oh… oops.


Yeah if they didn't have to go very far...

> It was exhausting to control and, according to program lead Ralph Mosher who was the designer and primary driver, operators could only drive the walking truck for a limited time.


Video: https://www.youtube.com/watch?v=ZMGCFLEYakM

The thing looks a little top heavy to me. I wonder if there are airbags :)

Also - external hydraulics seems like quite a limitation.


Between the old footage (minus the basic color), music, and narration this feels like it could be a scene from the twilight zone or a doctor who episode

Interesting, perhaps the inspiration behind the walking battle machines[0] in my favourite table top board game "Scythe"[1].

Interesting to see a real life Mech!

[0] - http://www.heavymetal.com/news/these-amazing-paintings-raise... [1] - https://stonemaiergames.com/games/scythe/


> It alternatively bore the name of "CAM", an acronym for "cybernetic anthropomorphous machine"

now that's a seriously cool name!

The pic makes me think of Strandbeests and derivatives: https://www.strandbeest.com/


It's the opposite of cool name. Anthropomorphous means resembling human being, this was nothing like human, it was quadruped.

That just makes it an incorrect name but not necessarily not a cool one.

The Death Star isn't a star but the name is cool.


It had four limbs that were directly controlled by a human's four limbs. That's a lot more anthropomorphic than any other existing vehicle.

They look vulnerable to snowspeeders.

That armor's too strong for blasters.

The reminds me of this video [1] where the guy builds these out of wood and paper and powers them with wind. Increadible.

[1] https://m.youtube.com/watch?v=LewVEF2B_pM


That thing looks amazing! A similar concept was these walking harvesters https://www.youtube.com/watch?v=2pJwDZXasKU

That was an interesting idea. Deere acquired them, and built a prototype.[1] But Ponsse's approach[2] was more practical. The dual chassis articulated 8-wheel system is unusual but very effective.

A big problem with those heavy legged machines is very low road speed. The Ponsse machine is much faster when you aren't in difficult terrain, so you can get it to the job under its own power. Deere copied that and now makes those, too.

[1] https://agmetalminer.com/mmwp/wp-content/uploads/2012/04/wal...

[2] https://www.youtube.com/watch?v=PE1f1GydafQ


The Ponsse machine: it would be cool if they could recover the gravitational potential energy of the tree for the cutting operations, say by storing the energy in a flywheel

One of the advantages of a legged machine is its smaller footprint on the often very vulnerable (at least in boreal forests) forest floor. Wheels and tracks tear and dig into the soil and cause a lot of damage that can take years to heal.

There was a walking logging machine developed by Finnish company Plustech Oy in the mid-nineties and later bought by John Deere who tried to commercialize it but it didn't meet much success.

"[It] was designed with sensitive or by other means harder to reach terrain with minimum impact on the environment in mind."

https://www.youtube.com/watch?v=2pJwDZXasKU

http://www.theoldrobots.com/Walking-Robot2.html

http://www.unusuallocomotion.com/pages/museums/museum-of-lus... (Near the bottom)



I’ve always wondered how much money consultants made off the “cybernetics” buzzword in the 60s and 70s.

Was it a buzzword at the time? Stuff like this is actually closer to the original definition--very roughly, a system that uses continuous feedback loops to keep itself in a stable, functioning state--than to the way we use the word now.

Why are these walking machines always so slow? I think Neal Stephenson described a skateboard with hundreds of tiny feet, moving faster than wheels on any surface. I want to see that.

Progress takes time, and ambulatory motion is a finicky, delicate nightmare that only seems simple in humans and animals thanks to millions of years of evolution. CAM was created in the 1960s[0], but BigDog[1] and other recent Boston Dynamics quadrupeds that aren't designed for human passengers are much faster[2], but still awkward.

Also remember these are expensive prototypes designed to test things other than speed. Making them faster only guarantees that they break themselves harder when they run into a wall or trip over a cliff or something.

[0]https://www.youtube.com/watch?v=ZMGCFLEYakM

[1]https://www.youtube.com/watch?v=W1czBcnX1Ww

[2]https://www.youtube.com/watch?v=3OKZ_n8QW4w

>I think Neal Stephenson described a skateboard with hundreds of tiny feet, moving faster than wheels on any surface. I want to see that.

I am neither Neal Stephenson nor a physicist but I'm pretty sure the friction caused by the increased surface contact of "hundreds of tiny feet" versus wheels would still make the wheels faster and more efficient. Geckos and things can stick to vertical surfaces and move quickly because they're lightweight. Anything capable of sticking to an arbitrary surface and supporting human weight is going to be slow. And if moving downhill, wheels don't really need to do work - gravity does the work. Something with "hundreds of tiny feet" either has to slide or "walk."


> Something with "hundreds of tiny feet" either has to slide or "walk."

IIRC the Neal Stephenson version was tiny feet mounted on wheels - not walking or sliding but rolling. But better than rubber pneumatics at evening out curbs or other obstacles.


To me, I envisioned them as being a wheel made with a bunch of shock-absorber leg-like "spokes" with feet.

They seemed similar to old-tech Pedrail wheels:

https://en.wikipedia.org/wiki/Pedrail_wheel

I think something like them has been made - but instead of the spokes ending in "feet" - they instead were encircled by a rubber deformable tread; Tweels are another interesting style of this technology, and simpler mechanically:

https://en.wikipedia.org/wiki/Tweel


Sounds more like legs than feet...

For the record, Stephenson's skateboard smartwheels used a bunch of little pads mounted on telescoping spokes instead of a solid wheel. The board had a terrain-scanning laser mounted in the nose, and pulled the spokes in and out so that the wheel was perfectly shaped for the contours of whatever it was about to roll over; you can ride a smartwheel board down stairs as smoothly as you would a ramp. But they were still very much wheels. There was no walking motion involved.

I disagree. Centipedes and other many-legged critters can move very fast. Every leg is just a tiny lightweight thing with its own balanced dynamics, just a part of a wheel basically. First thing that came to my mind is a plastic brush with flexible back. If you move a rollers along the back the bristles extend downwards and move backwards in a walking motion, just like a sea urchin does. Scale that and you have a walking machine moving amazingly fast. Tiny bristles yield and conform to any surface.

You're not taking inertia and friction into account.

Centipedes appear to move very fast because they're small relative to us and lightweight - at scale, they don't move very fast. But if you take a centipede and scale it up to the size of a train (square cube law notwithstanding), it would just be a very slow train-sized centipede. To make it faster, you would have to make its legs longer, because legs are levers.


>Centipedes and other many-legged critters can move very fast.

Can they?

Humans have legs too, but they aren't very fast at all. I can easily outrun any human on my bicycle.

The main advantage for legs is dealing with rough terrain. I can hike over/through extremely rough terrain that I would never dream of taking a bike on. But for speed, nothing beats wheels on a smooth surface. You can't have both.


CWU's "whegs" are much in this vein: http://biorobots.case.edu/projects/whegs/

I could also see Brook's Genghis platform being extended to "centipede" multiples of legs and probably working well.

https://en.wikipedia.org/wiki/Genghis_Robot

EDIT: fixed mistake - put "Dante" but meant Genghis...


I found such a brush with flexible back (from a broken sauna brush). Amazing discovery: If you move the said roller forward along the back the bristles make the walking movement backwards. Need to make room for the Nobel prize now.

Stephenson's skateboard smartwheels were still wheels. They moved by rolling, not running. They just used their extendable pads to roll smoothly over rough terrain.

It had force feedback. Did it have any sort of protection in the case of rapid movement of tge legs?

Oh wow that typo went by unnoticed

I personally don't know if it did - I've never seen any information on operator safety of this machine, though I have read a lot about it and Ralph Mosher over the years.

He's also the guy behind the GE Hardiman exoskeleton:

https://en.wikipedia.org/wiki/Hardiman

...which was another system much like this truck. For that one - and I don't know how true it was - it was said that it could get into a feedback loop that could cause injuries to the user, but it wasn't clear if that was for the entire "suit" or for the mounted arm test assembly (they had built a single arm up to the shoulder and mounted it on a stand to test it separately from the "suit").

GE was heavily into this field of what they called "man amplifiers" - several of the later ones spearheaded by Mosher:

http://cyberneticzoo.com/man-amplifiers/1953-g-e-o-man-manip...

http://cyberneticzoo.com/man-amplifiers/1958-9-ge-handyman-r...

http://cyberneticzoo.com/walking-machines/1962-64-ge-pedipul...

http://cyberneticzoo.com/man-amplifiers/1965-g-e-lifting-boo...

http://cyberneticzoo.com/teleoperators/1969-g-e-man-mate-ind...

This was also the same time period as Hughes Aircraft's MOBOT systems - which were meant as remote manipulators for working in nuclear research, underwater repaid, and similar harsh environments:

http://cyberneticzoo.com/teleoperators/1959-mobot-1-hughes-a...

http://cyberneticzoo.com/teleoperators/1962-underwater-mobot...

http://cyberneticzoo.com/teleoperators/1963c-mobot-hughes-ai...

http://cyberneticzoo.com/teleoperators/1964-%e2%80%93-mobot-...

What we now know as "ROVs" (and UAVs, UGVs, and other names) all stem from a lot of this research back then (and earlier). Hughes' system was actually kind of innovative; there is a paper you can dig up (in fact, I think it's linked on one of the above articles) that detail how these systems were controlled:

They needed to control a lot of various servo and other actuators, but they couldn't use a long cable with tons of wires - they needed the cable to be flexible and not weigh a ton (which is still an issue today). But they didn't have the modern electronics we have today to accomplish it. What they ended up doing was using a synchronized motor-switching system that had two rotary switches driven by synchronous motors on both ends. There was a mechanism in place to keep them "in sync" so that the switch connected on one end would be the same as on the other end. The controller at the transmitter would thus be connected to the proper motor or actuator at the other end, many times a second, and the umbilical only needed a few wires (plus wires for power - which IIRC was AC and served as part of the sync system - plus wires for video feeds). In a way, it was an early form of electro-mechanical multiplexing.

It quickly gave way to more electronic methods (likely things more closely resembling serial PPM, used in hobby radio control systems - once reliable transistors became common; while such a thing could have been done with vacuum tubes, it wouldn't have been reliable on such machines and environments) - but reading that paper was an interesting insight on how to do things differently! In fact, it's similar in scope to how Westinghouse did something similar for telemetry monitoring and control back in the 1920s and 30s - and showcased in their "robot" Electro - by using sounds and tuning forks, in a similar was to electrical coherer systems - they could control relays remotely using tones sent down telephone lines. Why Hughes didn't use a similar approach is a mystery - could have been patent reasons, or reliability issues or something of that nature.

It's a fascinating period of history in robotics, control theory, computation, and more in my opinion (also at this time, you have a large interest going on in artificial intelligence - such as it was in those years - and deep behind the scenes, there was certain interest in so-called "artificial neurons" - hardware implementations of the McCulloch-Pitts neuron model. At one time, around 1956 IIRC, RCA even made a "frog's retina" model using similar models - when you look at it and how it was wired and works - it really looks like a CNN made of hardware for image recognition...


Hey, that's very cool. Thank you!

Discontinued when they realized the enemy could just fly around it’s feet with a tripwire. ;) Inspiration for the AT-AT?

Is this better than a Driving Truck?

Metal Gear? : )



Applications are open for YC Winter 2020

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: