Quadruped Robot/Walker in KSP

[Jollyfellow]

One of the first things I tried to do with the Breaking Ground is make a quadrupedal robot in a similar fashion to BostonDynamics SpotMini.
It... works.
Barely

This version has a "Stand" and "Walk" function, with work-in-progress "Sprint" function. It's almost uncontrollable without reaction wheels, but that's likely due to poor programming on my part, of which was a pain to do.
The main issue with this style of quadruped that I found is that the servos don't move quick enough to get any decent speed from it, though it does still work.

Here is a video showing it walking:
https://youtu.be/QLtOllMOOuU

The craft file can be found here:
https://steamcommunity.com/sharedfiles/filedetails/?id=1769266564

If anyone can make it work better, I'd love to see it!

[Geschosskopf]

16 hours ago, Jollyfellow said:

If anyone can make it work better, I'd love to see it!

That's working marvelously.  What's to improve?

The speed of walking robots is limited by 3 things:

1. The speed at which the hinges/servos move, which basically sets the pace of the robot's gait (steps per minute).  You can set "traverse rate" to max but there's an upper limit you can't exceed.  And the more weight the part is moving, the slower it goes.
2. The length of the stride.  The longer the leg, the more ground it covers per movement of the leg.  So for the same hinge/servo speed, a longer leg (provided it actually takes bigger steps) will result in a faster robot.  However, longer legs weigh more so eventually the speed lost by the hinge/servo counters the speed gain from the long leg.
3. The amount of traction between the feet and the ground.  If the foot slides backwards while moving the body of the robot forward, it shortens the effective length of the step, thus reducing speed.

In  your example here, you seem to have the joints moving as fast as possible, they're taking reasonably long steps, and I saw no slipping of the feet at all, even on the runway, which is more slippery than the ground beside it.  So I think you're about as fast as you can go.

That's the thing.   Due especially to #`1 and #2, walking robots are SLOW in KSP.  It's just a fact of life.  More like AT-ATs than Voltron subcomponents.

[Foxster]

Interested in the ground slippage thing cos my attempts at a walker aren't going well due in part to this. 

Next thing I'm going to try is landing struts for feet because they are much more surface sticky than other parts. 

[GluttonyReaper]

53 minutes ago, Foxster said:

Interested in the ground slippage thing cos my attempts at a walker aren't going well due in part to this. 

Next thing I'm going to try is landing struts for feet because they are much more surface sticky than other parts. 

One thing that seems to help (at least for slow walkers) is having pistons as "feet", and having them extending them down into the ground in time with each 'step'. 

[FleshJeb]

7 hours ago, Geschosskopf said:

The speed at which the hinges/servos move, which basically sets the pace of the robot's gait (steps per minute).  You can set "traverse rate" to max but there's an upper limit you can't exceed. 

You can lever unpowered servos faster with pistons. The extension rates are 1,2,5, & 10 m/s. It's mathematically possible to construct "musculature" that will exceed the native rotation rate.

I was trying to do an ornithopter last night, but I got the rigging wrong and there was too much joint expansion. However, I think I've figured out an assembly that will minimize this. I'm hoping it's possible to slam these suckers around in a stable manner.

[sturmhauke]

On 6/13/2019 at 3:34 PM, Jollyfellow said:

One of the first things I tried to do with the Breaking Ground is make a quadrupedal robot in a similar fashion to BostonDynamics SpotMini.
It... works.
Barely

I made a quad walker too, but yours is more stable. Mine had greater articulation (knees, elbows, and ankles), but proved difficult to time correctly.

[Geschosskopf]

7 hours ago, Foxster said:

Interested in the ground slippage thing cos my attempts at a walker aren't going well due in part to this. 

Next thing I'm going to try is landing struts for feet because they are much more surface sticky than other parts. 

Lander legs have the lest slippage but you need to kill their springs and max their damping or bad things happen.  And even then, they're kinda wonky.  As an alternative, you can increase the foot's surface area.  Even relatively slippery parts get reasonably good grip if you have a big enough shoe size.

 

1 hour ago, FleshJeb said:

You can lever unpowered servos faster with pistons. The extension rates are 1,2,5, & 10 m/s. It's mathematically possible to construct "musculature" that will exceed the native rotation rate.

I was trying to do an ornithopter last night, but I got the rigging wrong and there was too much joint expansion. However, I think I've figured out an assembly that will minimize this. I'm hoping it's possible to slam these suckers around in a stable manner.

My fastest walkers have used pistons straight ahead and astern in a push-pull configuration to move the body forward, then hinges to raise the feet off the ground while the pistons reset.  Meanwhile, another identical set of pistons takes the next inchworm step.  Very ungainly but it works.  Still, even this is limited by the speed of piston extension and that's only about 5m/s.  In this case, there's no "cosine loss" from pushing an otherwise unpowered leg, but there's no gain from the lateral length of the leg, either.  Still, there's only a small structural panel on the end of the each piston to prevent slippage.  Thus, the mass the pistons are moving is reduced to the main body of the walker, and the other set of legs resetting during the step.  This is less mass than the body of the rover plus longer legs with multiple joints, so the piston motion is faster.

So, at the bottom line, complexity = slow for walkers.  And walkers are already slow to begin with.  Thus, if you're after speed, use rover wheels.  But if you're after smooth, walking motions, then knock yourself out with multi-jointed walkers.

[Guest]

Nice!

[Snark]

Moving to Breaking Ground Discussion.

[Klapaucius]

I invite you to @Kergarin's challenge:

 

 

On 6/15/2019 at 3:41 AM, Foxster said:

Next thing I'm going to try is landing struts for feet because they are much more surface sticky than other parts. 

Right now the OP's walker has only two joints--a hip and a knee.  It is missing an "ankle".  Also, there are the joints  around the ball of the foot that allows humans to stand on their toes.  Consulting with my veterinarian wife, she confirmed that dogs and cats are walking on their toes.

 

Check out this link and scroll down to the dog skeletal diagram.  You see how the ankle is far higher than on a human and the metatarsus is vertical.

https://www.allthingsdogs.com/dog-anatomy/

 

 When I made my frog (which granted is a different animal in more ways than one) I used an I-Beam for the foot and hinged it so it sat flat at rest.  I don't know if KSP physics models this, but it does increase surface area. The I-beam also has a slightly higher impact resistance than the DLC parts, so it makes a good "shoe".

 

PS: @Jollyfellow--Any chance you could put that craft on something else like Dropbox, Google Docs or KerbalX so we don't have to deal with Steam Workshop?

Edited June 18, 2019 by Klapaucius