Help with finding solutions for getting an AT-AT walk cycle like in the movies.

[BrickBear]

On 12/15/2024 at 5:16 PM, Divitis said:

Can we see it moving?

Pretty pleaseeeeeee

New version moving, lossy gif. No legs yet. 

 

Edited December 26, 2024 by BrickBear

[BrickBear]

Some good news, i’ve begun adding the legs and implemented various features to counteract some problems I was having.

One problem I was facing was when placed the legs would slide (as if doing the splits). I had this same issue on the previous one which is why I remade the chassis in the first place, I thought it was a suspension issue but To fix this I merely needed another 2x1 beam to be inserted onto the existing cam shaft to act as a stopper which could have been done on the old chassis. But nevermind, this chassis is stronger so it’s progress either way.

Referring to @Nelson suggestion to take the weight off of the back leg diagonally opposite to the current leg lifted up off of the ground I have found that now without the “splits” motion occuring for my model that this is a natural byproduct of my mechanism design anyway. As a result upon adding the other legs it stood perfectly even with one leg raised.
 

I just hope the torque of the xl motor is enough to move everything once the bodywork is placed. I’ve defrictioned joints, modified the suspension all to reduce resistance on the motor. I believe the XL motor can take up to 14nm according to wikipedia 9.8nm is roughly enough for a 1kg weight to be lifted a metre. 
 

I’d be interested to know how much resistance towards the motor the hard suspension provides in terms of weight. (I guess that would be how much weight the spring can hold at full length under load?) 

 

[Nelson]

@BrickBear I'm guessing you've already referenced this site, but it's always good to put it out there for folks to check out.

https://www.philohome.com/motors/motorcomp.htm

I'm not well versed in how electric motor power is rated, but I can say that, in my model with two large Powered-Up motors, the weak points are the gears, axles, and general flex of parts, not the power. I really like the gif of your chassis. I'll be jealous if you beat me to a full working model. I will say that I had a very early chassis-leg version working a few years back that used springs in a similar fashion. Unfortunately (as seems to be the Achilles heel of all my designs), the springs could not deal with the weight once the full body cladding was installed. Perhaps your bodywork will be light enough.

I replaced the broken gears on my most recent design, but couldn't even get it to take a single step the next time I fired it up. The gears jumped immediately. Weird how a design can work one day and fail the next. I've come to realize that, with the margin of success and failure so slim, wear and fatigue on parts can make a big difference. IE: In my model's drivetrain, I have a fairly long run of axles that connect the motors (slung underneath) with the final leg mechanisms. The flex of this drivetrain can get pretty severe. There are times when two legs can almost be on the same cycle if one binds up. The power continues to drive the other legs until the flex of the drivetrain reaches its limit and everything grinds to a halt. I've noticed this effect grows worse over time as the axles are flexed over and over again. Eventually, they lose a good portion of their integrity even though they don't break. It may be worth rethinking my overall layout in order to get the power output and the leg mechanisms connected more directly.

I think I'm going to give this "slide piston" version another go. I want to add some rollers for the cams to contact against. I think the friction between the cams and their contact points is creating too much resistance now that I have some rubber bands assisting the lift. It's interesting that I have rubber bands pulling up that need to be overcome on the downward motion while you have springs pushing down that need to be overcome on the upward motion. It's definitely a tough battle to fight from either direction!

8 hours ago, BrickBear said:

New version moving, lossy gif. No legs yet. 

 

I'm concerned that your leg lift times overlap too much. It appears that the front leg hasn't completely come down before the rear leg is coming up. What if you left a bit of gap between the rotating liftarms and the lever? Maybe use  a 4-length and 3-length liftarm instead of a 3 and 2, but have some space where there's no contact?

[BrickBear]

@Nelson

First of all may the best man win.
 

You raise a good point about the leg raising I could try adjusting the positions of the cams by disengaging the middle. 
 

And yeah i’ve checked out Philo’s home been using it since I was a kid. Regarding the weight of your model, do you think it might be worth sacrificing certain levels of detail for functionality? Like maybe you could get some prints or stickers for detail on larger plates? 

[BrickBear]

Ok so I also introduced stoppers to the front. This has significantly improved balance as the legs are nwo limited to the movement they should have been. 

Also strengthened the central pivot for steering. Everything has been reinforced basically.

Need to see if it walks now.

 

[BrickBear]

Whose a good boy? I realised I couldn’t make it walk without the head being added as that’s the heaviest part.

(Note the head seems bigger here, it’s just perspective)

Lifting the feet with head acting as a counterweight.

 

What i’ve noticed is that the feet do still slip a little on the ground with the head added. But the springs are really good at holding the weight. I also have new batteries which really upped the power of the xl motor. 
 

My main problem now is the head function, I need to have the head rotate based on the movement of the legs so it remains an effective counterbalance.

Still, the model has never been so stable! 

 

[LegoTT]

Hello, why do it have a small turntable on the top ?

[BrickBear]

5 hours ago, LegoTT said:

Hello, why do it have a small turntable on the top ?

Thats just where the steering pivot point is, a turntable is a stronger pivot piece than an axle or pin.

[BrickBear]

A new year, perhaps a time for new solutions. Here’s to achieving our AT-AT ambitions in 2025 

[Nelson]

I decided to try a different drivetrain configuration. Instead of the rather long path of having the motors positioned below the body going to a worm gear, then up through the body's center into another gear reduction, then along the length of each forward and rear half (whew!), I put the two large motors at the top of the model, directly driving a 40-tooth gear via worms. I was concerned this would make the model too top-heavy, but with the battery box moved below, the weight distribution is about the same. The first run was promising, but as expected, the cams worked themselves free of their drive gears pretty quickly. Still, this is promising. I can solve that issue by adding supports around the cams, but then I run into problems where the width of the internal mechanisms are outside the boundaries of the bodywork in the final model. I'm working on a new design that will deal with that and will hopefully have it tested this weekend.

You can see when the rear cams become disengaged in this video. You can also see that I need to add a bit of assist (probably a mildly tensioned rubber band) to get the rear legs down.

 

[BrickBear]

5 minutes ago, Nelson said:

I decided to try a different drivetrain configuration. Instead of the rather long path of having the motors positioned below the body going to a worm gear, then up through the body's center into another gear reduction, then along the length of each forward and rear half (whew!), I put the two large motors at the top of the model, directly driving a 40-tooth gear via worms. I was concerned this would make the model too top-heavy, but with the battery box moved below, the weight distribution is about the same. The first run was promising, but as expected, the cams worked themselves free of their drive gears pretty quickly. Still, this is promising. I can solve that issue by adding supports around the cams, but then I run into problems where the width of the internal mechanisms are outside the boundaries of the bodywork in the final model. I'm working on a new design that will deal with that and will hopefully have it tested this weekend.

You can see when the rear cams become disengaged in this video. You can also see that I need to add a bit of assist (probably a mildly tensioned rubber band) to get the rear legs down.

 

Looks like a brilliant start, I have to admit I’m somewhat jelly of the stability of yours. 

[Nelson]

@BrickBear I believe having the legs laterally stiff throughout their cycle is essential, at least in my design concept. I gave up on the bending knees early on in favor of stability. I admire your efforts to include that functionality. If you succeed, I will certainly consider adding it back. My hope is to get the appearance of bending knees by having the upper portion of the leg be a sort of hinged box around the straight liftarm inside. I'm not exactly sure how I'm going to control the movement of that upper leg. I'll burn that bridge when I come to it.

 

Edited January 2 by Nelson

[oracid]

The AT-AT has a gait that I would call the camel gait. One side, then the other side.
It is this gait that marks the recognizable style of the AT-AT.
I invite you to watch these 2 videos.

https://youtu.be/GUsOouwjsL4?si=c3JUcEu3YiOGfX_W

https://youtu.be/bv_cE58WMzs?si=be5C6v1HG8mwgW19

 

[BrickBear]

12 hours ago, Nelson said:

@BrickBear I believe having the legs laterally stiff throughout their cycle is essential, at least in my design concept. I gave up on the bending knees early on in favor of stability. I admire your efforts to include that functionality. If you succeed, I will certainly consider adding it back. My hope is to get the appearance of bending knees by having the upper portion of the leg be a sort of hinged box around the straight liftarm inside. I'm not exactly sure how I'm going to control the movement of that upper leg. I'll burn that bridge when I come to it.

 

Ooh that is clever. 
 

 

49 minutes ago, oracid said:

The AT-AT has a gait that I would call the camel gait. One side, then the other side.
It is this gait that marks the recognizable style of the AT-AT.
I invite you to watch these 2 videos.

https://youtu.be/GUsOouwjsL4?si=c3JUcEu3YiOGfX_W

https://youtu.be/bv_cE58WMzs?si=be5C6v1HG8mwgW19

 

I’ve watched the first an unhealthy amount of times. It’s interesting how the camel lifts at the shoulder at its front like Nelsons. It’s definitely the same gait (albiet back leg first then front leg)

[oracid]

5 hours ago, BrickBear said:

It’s definitely the same gait

Are you sure ! ?

At 0'20" in the first video : Left Front Leg  -  Left Back Leg  -  Right Front Leg  -  Right Back Leg

Edited January 3 by oracid

[BrickBear]

6 hours ago, oracid said:

Are you sure ! ?

At 0'20" in the first video : Left Front Leg  -  Left Back Leg  -  Right Front Leg  -  Right Back Leg

And then I said but back leg first, the camels is reversed, I acknowledged that.

[LegoTT]

About stability and body weight reduction, have you considered to go on a slighty bigger scale and put one motor or small battery box

https://www.bricklink.com/v2/catalog/catalogitem.page?P=4760c01#T=S&O={"iconly":0}

IN each leg (weight touching the ground instead being in the body), going up trough the leg with cardans at high rotation speed low torque and massive worm gear reduction just before the walking mechanism ?

According philohome :

https://www.philohome.com/motors/motorcomp.htm

Three (3) 2838 motors

https://www.bricklink.com/v2/catalog/catalogitem.page?P=2838c01&name=Electric, Motor 9V 5 x 4 x 2 1/3 with Black Axle and Dark Gray Base&category=[Electric, Motor]#T=S&O={"iconly":0}

(48 gram each) give as much power than a PF XL motor and we can put one batterybox in the 4st leg (about 64 gram filled with the battery)

 

I don't know if you can have full power of three 2838 motor with only one 9v battery tough, but I think that this battery box doesn't have current limitation so it will rely on the exact brand of 9v battery, some of them are engineered to have a smaller internal resistance.

 

 

[BrickBear]

6 hours ago, LegoTT said:

About stability and body weight reduction, have you considered to go on a slighty bigger scale and put one motor or small battery box

https://www.bricklink.com/v2/catalog/catalogitem.page?P=4760c01#T=S&O={"iconly":0}

IN each leg (weight touching the ground instead being in the body), going up trough the leg with cardans at high rotation speed low torque and massive worm gear reduction just before the walking mechanism ?

According philohome :

https://www.philohome.com/motors/motorcomp.htm

Three (3) 2838 motors

https://www.bricklink.com/v2/catalog/catalogitem.page?P=2838c01&name=Electric, Motor 9V 5 x 4 x 2 1/3 with Black Axle and Dark Gray Base&category=[Electric, Motor]#T=S&O={"iconly":0}

(48 gram each) give as much power than a PF XL motor and we can put one batterybox in the 4st leg (about 64 gram filled with the battery)

 

I don't know if you can have full power of three 2838 motor with only one 9v battery tough, but I think that this battery box doesn't have current limitation so it will rely on the exact brand of 9v battery, some of them are engineered to have a smaller internal resistance.

 

 

So I actually came up with and implemented a much simpler solution for weight distribution on the front feet this morning. (The back feet lift fine already because the head leans the weight away from them)
 

It’s basically all about the heads attachment system. 
 

So above the front two legs “shoulders” are a sort of see saw mechanism. When one front leg lifts up it causes the seesaw to rise on the side of the raised leg. The head which is connected to an axle on the seesaw then rotates toward the front leg that is not raised because of weight and gravity. Because the see saw is moved directly by the legs, the weight on the seesaw created by the head is transferred to the leg that is not raised and so stability is achieved. It’s so simple and such a relief to have achieved this. 
 

Also @Nelson you might find this solution interesting.

Edited January 4 by BrickBear

[Nelson]

Another chassis update. This one eliminates pretty much all of the slack and flex in the drivetrain with two motors directly driving their respective cranks. Still a ways to go, but this feels like a major step forward (pun intended).

Update: I’ve been tweaking and tinkering for a couple hours now. It’s the first design I’ve built that can keep walking without the mechanisms self destructing in a few steps. It’s revealing other weak points that weren’t apparent before. I think I’ll need to redesign the legs. There’s an area in the shin that eventually separates after about two or three minutes of walking. I’m still really thrilled with this progress though.

Edited January 11 by Nelson

[BrickBear]

8 hours ago, Nelson said:

Another chassis update. This one eliminates pretty much all of the slack and flex in the drivetrain with two motors directly driving their respective cranks. Still a ways to go, but this feels like a major step forward (pun intended).

Update: I’ve been tweaking and tinkering for a couple hours now. It’s the first design I’ve built that can keep walking without the mechanisms self destructing in a few steps. It’s revealing other weak points that weren’t apparent before. I think I’ll need to redesign the legs. There’s an area in the shin that eventually separates after about two or three minutes of walking. I’m still really thrilled with this progress though.

Brilliant. Looks very exciting.

[2GodBDGlory]

Impressive progress!

[Nelson]

If there's one thing I know how to do, it's take a design that's functioning at 70% and make it function at 25%. All my attempts at correcting the issues with the recently posted design have yielded worse results.

I think the reason the rear legs are a bit too forward in timing is because I have to use 44809 pin connectors everywhere. It causes the geometry of the rotational centers of the slide piston mechanism to be a stud off center in some areas. I don't know why LEGO doesn't make a pin with pin hole (15100) without friction ridges. That would be so useful! With the front and rear mechanisms being flipped mirror images of each other, that subtle offset compounds to make the lifting motion a bit late on the front and a bit early on the rear (while the cams remain timed correctly). It throws everything off just enough to be annoying.

I'm rambling.

I'm considering going back to elements of an earlier design but keeping the new motor/drive setup. For a long time I thought there were other issues at play, but now I see just how much of my failures were caused by the flex and power loss in the drivetrain when I had all those axles connecting the motors to the cranks. Maybe by combining the new with the old, I'll get it right.

Edited January 11 by Nelson

[Appie]

1 hour ago, Nelson said:

I don't know why LEGO doesn't make a pin with pin hole (15100) without friction ridges. That would be so useful!

Perhaps you already considered this new part from 2024, but just in case: 

https://www.bricklink.com/catalogPG.asp?P=5713&colorID=86

I have no idea if it can fit in your design, but at least you'd sort of have a frictionless 15100 like this with some half bushes at the rear? 

[2GodBDGlory]

Or you could use parts 27940 or 22961 with support/blocking on the other side:

[Nelson]

I made some adjustments to the cam contact areas and got more lift. This helps the inside "toe" clear the ankle stabilizer. I may make the toes hinged so they clear better and look more accurate. (I had that feature on an early model.) I'll just have to see if it compromises stability. The cadence is still a bit off. I'm pretty sure this is a result of the mechanism's geometry being imperfect and the limited range of the ankle joint. The rear foot can't flatten out when the leg comes down because the ankle only has a flex of a few degrees. This is a fine line to get right. I could try to get a bit more flexibility range, but then risk the walker falling back on its rear. (I believe @BrickBear was experiencing this phenomenon.)

I think the next step is to try and add some body panels and a rudimentary head so I can see how the design handles the weight and balance changes. 

There's also the issue of the legs which begin to split apart after a while. I'm bummed about that. I spent a massive amount of time getting these legs designed (with strength in mind) and I feel like the options of build styles I have for getting better longevity will compromise the ascetic. But hey. That's part of the challenge.

 

 

Edited January 15 by Nelson