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

[BrickBear]

 

Hello, this is my first proper thread so forgive me if it’s a little amateurish in layout. With that said i’d like to begin by saying i’ve been working on this project on and off for a few years to get my 75054 AT-AT walking, having tried out the mechanism for the official LEGO walking AT-AT i’ve found that it is not stable enough nor satisfying. 

 

Since then I have been through multiple iterations of trying to get each leg moving but with only one off the ground at a time. 

 

As you can see in the pic from DK’s StarWars Vehicles book the mechanism for connecting and moving the lower leg relative to the upper leg seems to be similarly joined enough that it can be replicated using a linkage but making those linkages move in the correct sequence is the difficult part. The ankle on an AT-AT keeps three legs at a time level in the ground as well as lifts one other leg by moving it up when it takes a step. I have only ever found one person who has managed this in a non LEGO model so it’s possible:

 

https://www.youtube.com/watch?time_continue=3&v=34tgu7ULkeM&embeds_referring_euri=https%3A%2F%2Fwww.google.co.uk%2F&source_ve_path=Mjg2NjY&feature=emb_logo

 

That was 15 years ago.

 

Here are some of my attempts: 

 

These two images use two sets of half beams to raise, lower and flatten the beams holding the leg, allowing up down movement, connected to the lower half beams are some linkages to move the leg back and forth, the fault of this one is it was not smooth nor stable.

 

front

Back (blue and red beams provide strength)

This mechanism uses clutch gears going in a forward direction to move the bars down, half beams make them move up and down in turn, this reduced vertical space taken up, the failure of this is lack of connection points and strength as well as clutch gears probably not being able to support the weight.

This was the most promising one, strong too but was unfortunately not quite walking in the right motion, it was not fully stable either.

From further study I have noticed that the legs only bend in one direction so i’ve placed a stopper on each one to mimic this, I imagine this will add to the challenge. But i’ve seen people on eurobricks work magic with technic so I figured I should ask here for some help to find a solution. 

Edited December 2, 2023 by BrickBear

[allanp]

I think this will be tricky with only cranks and levers. If we look at each phase, the leg must lift up off the ground, move forwards and drop back down to the ground in 1/4 of a cycle, then slowly move back at a steady speed for the other 3/4 of the cycle. I think, to get that truly AT-AT motion, I'm thinking a cam arrangement might be a way to brute force the problem into submission. For example, maybe take three of https://www.bricklink.com/v2/catalog/catalogitem.page?id=922&idColor=3&ccName=4500483#T=S&C=3&O={"color":3,"iconly":0} to make a 3/4 circle shaped cam that pushes the leg down to the ground. Have the cam hidden up in the body and directly above the leg for strength. The leg would be spring loaded to lift off the ground, this only needs to be strong enough to raise the leg. Then the cam pushes it down only having to be strong enough to overcome the spring that raises the leg. Then the weight of the model is on the perfectly circular part of the rotating cam.

So this cam gives you a position in 1/4 of a cycle (leg up) while giving you another position in the other 3/4 (leg down). That would also be useful for the knee joint yes? You could use a tie rod to transfer the up/down motion of the hip to the bending motion of the knee, so down is knee straight and up is knee bent. Alternatively you could use another 3/4 cam in the leg if there's room for that, though you would also need to get a rotating shaft there also.

So now you just need the motion that bends the hip, making the leg move backwards (when the foot is down) at 1/3 the speed it moves forwards (when the foot is up). This is tricky. It might be possible if you have the forwards and backwards motion of the leg actuated via a rack and pinion, where the pinion is driven by a two position gearbox. In one gear the pinion rotates one way, and in the other position it rotates the other way at 1/3 the speed. The gearbox would be shifted via the up/down motion of the leg also. So in essence, everything is timed from that one 3/4 circle cam at the top of each leg. However, there is a chance for the gearbox to skip and lose it's position as the gear changes won't be instant. To reduce that you can hear the gearbox such that it spins faster and gear down afterwards, and make it so that it tries to move the leg beyond it's mechanical limit but is protected by a clutch gear, thus should help ensure the leg moves back and forth through the full distance. You can also simplify the gearbox by having not one, but two counter rotating input shafts at a 1:3 ratio, so then it's just a case of engaging each pinion of each leg to the correct shaft at the correct time. Is all that clear as mud?!

Of course there is a second option.... pneumatics. Instead of using the 3/4 cams to move the legs directly, you could use the cams to correctly time and control pneumatic valves in the correct sequence.

One valve moves the leg up/down and also use the same valve to bend/straighten the knee. Remember, leg up = knee bent, leg down = knee straight. You could also use the same valve to move the leg back and forth but I'd use a second valve so I can restrict the leg speed moving back to 1/3 in some way. Just make sure you have a beefy enough air compressor to keep up.

[BrickBear]

1 hour ago, allanp said:

Snip

That’s actually really helpful, I admit an image might help to visualise some of the rack and pinion + gearbox stuff but I can easily picture the 3/4 cam wheel pressing down on a beam that is pushed upward by a medium-hard shock absorber. Pnuematics would be cool (perhaps even linear actuators could work as I do have those)? You’ve definitely given me some food for thought.

 

edit: I’ve ordered some cams and hard shock absorbers. 
 

follow up question, the hips of an at-at from what I can tell move in a up down, level, down, (repeat) motion, to accommodate the legs bending (keep the at-at’s body level. Do you think it necessary to modify the cam to reflect this? Or should I consider that a relatively minor unnecessary aspect?

Edited December 2, 2023 by BrickBear
added questIon

[allanp]

4 hours ago, BrickBear said:

That’s actually really helpful, I admit an image might help to visualise some of the rack and pinion + gearbox stuff but I can easily picture the 3/4 cam wheel pressing down on a beam that is pushed upward by a medium-hard shock absorber. Pnuematics would be cool (perhaps even linear actuators could work as I do have those)? You’ve definitely given me some food for thought.

 

edit: I’ve ordered some cams and hard shock absorbers. 
 

follow up question, the hips of an at-at from what I can tell move in a up down, level, down, (repeat) motion, to accommodate the legs bending (keep the at-at’s body level. Do you think it necessary to modify the cam to reflect this? Or should I consider that a relatively minor unnecessary aspect?

Yes I know exactly what you mean, the hips do move back up again slightly to account for the leg moving through it's arc. It would be good if you could account for that somehow, maybe make a more complex shaped cam somehow, or move the cam slightly off centre, or mount the cam on an arm and move the whole cam up and down via a simple crank, or have the cam follower (the piece that touches the cam) also move slightly as the leg goes through it's arc. Not sure if that's needed, but I think it would definitely be more stable and look much more like an AT-AT.

For the rack and pinion, I'm thinking you would use it as the leg actuator (like the dump bed mechanism on the barcode truck from 1997), similar to a linear actuator but one that can move faster, but yes you could also use a linear actuator for this if you can get it to move fast enough. Then you use a gearbox to switch it's speed and direction. As you would need 4 gearboxes (one for each leg) you can simplify them by having two input shafts already spinning at a 3 to 1 ratio.

Another idea for the hip joint rotation, if you don't want a visible actuator of ANY kind (to look more like the AT-AT) you could use turntables like they did on the UCS version and drive them via a worm gear from the gearboxes. Actually I think this would give you the best speed as a rack might be too fast and a linear actuator too slow.

The pneumatic version might be simpler overall but be warned, getting the valves to move via the cams how you would want can be a bit finicky as the valves are quite stiff. You might want to use a second cam to move the valve the other way, however as the valve would stay in place at each side, you wouldn't need the cams to be any particular shape, so long as they move the valve at the right timing. Then to account for the leg sweep you can move the upper mounting point of the pneumatic up/down via a crank. If you do want to try it I would try to get those valves right first. 

As you have ordered the cams and shocks, would I be right in thinking you don't have that much Technic yet? Non of what I have said here is as easy to pull of as I have made it sound. But it sounds like a fantastic project to learn on and would be great if you can pull off. This will be a case of trial and error, breaking it up into smaller chunks and getting each part right, and also rigidity. You absolutely cannot make this too rigid. Any sloppyness or bending in the legs will undo all the work you have put into finely crafting each leg movement. 

Edited December 2, 2023 by allanp

[BrickBear]

59 minutes ago, allanp said:

Yes I know exactly what you mean, the hips do move back up again slightly to account for the leg moving through it's arc. It would be good if you could account for that somehow, maybe make a more complex shaped cam somehow, or move the cam slightly off centre, or mount the cam on an arm and move the whole cam up and down via a simple crank, or have the cam follower (the piece that touches the cam) also move slightly as the leg goes through it's arc. Not sure if that's needed, but I think it would definitely be more stable and look much more like an AT-AT.

For the rack and pinion, I'm thinking you would use it as the leg actuator (like the dump bed mechanism on the barcode truck from 1997), similar to a linear actuator but one that can move faster, but yes you could also use a linear actuator for this if you can get it to move fast enough. Then you use a gearbox to switch it's speed and direction. As you would need 4 gearboxes (one for each leg) you can simplify them by having two input shafts already spinning at a 3 to 1 ratio.

Another idea for the hip joint rotation, if you don't want a visible actuator of ANY kind (to look more like the AT-AT) you could use turntables like they did on the UCS version and drive them via a worm gear from the gearboxes. Actually I think this would give you the best speed as a rack might be too fast and a linear actuator too slow.

The pneumatic version might be simpler overall but be warned, getting the valves to move via the cams how you would want can be a bit finicky as the valves are quite stiff. You might want to use a second cam to move the valve the other way, however as the valve would stay in place at each side, you wouldn't need the cams to be any particular shape, so long as they move the valve at the right timing. Then to account for the leg sweep you can move the upper mounting point of the pneumatic up/down via a crank. If you do want to try it I would try to get those valves right first. 

As you have ordered the cams and shocks, would I be right in thinking you don't have that much Technic yet? Non of what I have said here is as easy to pull of as I have made it sound. But it sounds like a fantastic project to learn on and would be great if you can pull off. This will be a case of trial and error, breaking it up into smaller chunks and getting each part right, and also rigidity. You absolutely cannot make this too rigid. Any sloppyness or bending in the legs will undo all the work you have put into finely crafting each leg movement. 

I have a fair amount of technic for this project I think. I’ve taken apart 6x6 all terrain tow truck and airport fire truck for this as well as have a bunch of technic from the 80s onward (my cousins old lego) to use. The only parts I don’t have amazingly are shock absorbers or pnuematics. 
 

On another note I think I know what you mean, i’m gonna draw a little diagram but first I got to remember how to upload to flickr. Thankyou, 

[BrickBear]

18 hours ago, allanp said:

Yes I know exactly what you mean, the hips do move back up again slightly to account for the leg moving through it's arc. It would be good if you could account for that somehow, maybe make a more complex shaped cam somehow, or move the cam slightly off centre, or mount the cam on an arm and move the whole cam up and down via a simple crank, or have the cam follower (the piece that touches the cam) also move slightly as the leg goes through it's arc. Not sure if that's needed, but I think it would definitely be more stable and look much more like an AT-AT.

For the rack and pinion, I'm thinking you would use it as the leg actuator (like the dump bed mechanism on the barcode truck from 1997), similar to a linear actuator but one that can move faster, but yes you could also use a linear actuator for this if you can get it to move fast enough. Then you use a gearbox to switch it's speed and direction. As you would need 4 gearboxes (one for each leg) you can simplify them by having two input shafts already spinning at a 3 to 1 ratio.

 

Kinda like this?

 

 

[BrickBear]

Ankle Movement Mechanism, springs push beam down, cams push up, compact, simple, includes rotating middle to allow for gentle steering. Beige bevel gear on the right lower part of the mechanism is just me figuring out how to make legs move and not a required part of up down mechanism. 

 

 

[BrickBear]

I have got it to work and it has a kind of nice mechanical movement to it, I just need to refine the leg bending mechanism as you can see a simple yet fairly strong structure has been added to the top of the previously posted ankle mechanism, this joins one point of the leg with another and has a mechanism for forward and backward movement, as the ankle lifts up the leg bends due to the fact that one connection point is raised higher than the other and as it returns down it unbends (I do need to tweak things to male the leg bending a bit bendier), then it slides back, the solution turned out to be surprisingly simple and lightweight in the end however will need refinement to make it more compact and for the sliding part to be made more secure. 

 
Here is a gif as well: 

Edited January 1 by BrickBear

[2GodBDGlory]

Looks nice! A bit hard to wrap one's head around the mechanism, but it looks simple enough

[BrickBear]

22 hours ago, 2GodBDGlory said:

Looks nice! A bit hard to wrap one's head around the mechanism, but it looks simple enough

Thanks! I’m going to release instructions when I’m done. I just have to figure out the best software to do that as I’ll need to compress the springs in the mechanism to show things accurately. It’s genuinely exciting to see it come to fruition and to hear positive feedback too! 

[BrickBear]

 

Latest update, made the knee bending mechanism better, this is not the latest image but essentially what i’m trying to achieve. My current setup has the white quarter circle liftarm replaced with a 2x1 cylinder attached to one of the beams which either gets pressed down or lifted up by a stopper, however this while working is unsatisfactory as its not as strong a connection. The problem with the white liftarm was that the rotation movement was ever so slightly too far for the leg to go down properly, so the quest continues… 

 

 

 

Edited January 21 by BrickBear

[BrickBear]

I think this is the final leg iteration, it bends adequately, the “hip” raises up and down completely and the connections are fairly strong, the final bonus features are that I managed to slim the leg down by a whole technic bar so now each leg will be the same thickness as the original 75054 Lego AT-AT, the dish at the top rotates and the grey bricks with technic holes move up down like in the actual actuators in the AT-AT walker. When the other three legs are attached it will be the day walking tests commence, hopefully due to the hard suspension used on the hips all four legs will be able to support the weight of a battery box and motor. I should be able to implement mild steering too and with plenty of space for mini figs owing to the compact gearing.

[vascolp]

This seems very cool! I look forward to see it walking!

[BrickBear]

1 hour ago, vascolp said:

This seems very cool! I look forward to see it walking!

Thanks! Hopefully it’ll work!

[2GodBDGlory]

Great job persevering to get such a complicated linkage system working!

I look forward to seeing it in a model!

[BrickBear]

20 minutes ago, 2GodBDGlory said:

Great job persevering to get such a complicated linkage system working!

I look forward to seeing it in a model!

Thanks! 10 years of working on walking mechanisms have led to this moment. Hoping that it’ll work!

[BrickBear]

Okay, that was not the final iteration, I discovered a mistake in the placement of the leg on the ankle, the leg is three beams wide and the connection should have been in the centre, however I hadn’t done that so when it came to testing all four legs, there was a point where a leg would be too long. Likewise I have to work on balancing it so it can be stable on just three legs. 

[BrickBear]

Much better, as you can see the leg rotates on the centre beam and the mechanism for pushing and pulling the leg is placed inside, this has allowed me to bring the legs together, which i assume will make it more stable as the centre of gravity from a raised foot is closer to the other feet than it was before.

The last remaining problem is stopping the knee from bending when the leg is pushed backwards and it succumbs to the weight of the walker, I plan to introduce a stopper at the top of the leg to prevent this from happening so that when the leg faces back, the beams that go up and down parallel to each other to cause the bending can only be straight.

 

Incidentally I believe this new system where the leg is connected both inside and out will reduce wobble because it initially had a terrible lean towards whichever leg was lifted up.
 

This has sacrificed my ability to hide stuff on the rear of the leg but for the most part it is possible to cover all this up.

Edited February 27 by BrickBear

[2GodBDGlory]

Thanks for the updates! I look forward to seeing the result

[BrickBear]

On 2/27/2024 at 1:18 PM, 2GodBDGlory said:

Thanks for the updates! I look forward to seeing the result

Update time, ordered some new bevel gears from bricklink. The good thing about these ones is they have three connecting points within them much like the old spur gears. This means I can connect it directly to the legs and simplify the bending mechanism greatly. The one downside is due to its size the gear has to be attached half a beam away, this allows the liftarm connected to the orange cam to not bump into it. However the connection remains strong which is great and the leg bends well.
 

Edited March 10 by BrickBear

[BrickBear]

All four legs added, Struggling a little with its balance despite it having three feet on ground 

[BrickBear]

Ok I had to rework the gears of the legs, unfortunately the difference between the 20 tooth bevel gear and the 28 tooth bevel gear caused the legs to bend too far causing strain on the mechanism and knee stopper, so I changed it to 40 tooth straight gear and the 8 tooth straight gear but that was too little bend. I’ve now landed on 36 tooth and 12 tooth combination which is perfect and I sorted the balancing issue with the use of the head movement where it moves to the side of the body that does not have its leg lifted, this adds more weight to the limbs touching the ground. 
 

 

[BrickBear]

Akk I’ve just noticed, the weight at the front is a bit much for the position of the springs, gonna need to adjust that.

[Divitis]

Ao cool to see this evolve, please keep the updates coming.

What kind of body are you testing it with?

Are you 'animating' an existing one or working on it / tuning it in parallel with the legs?

[BrickBear]

2 hours ago, Divitis said:

Ao cool to see this evolve, please keep the updates coming.

What kind of body are you testing it with?

Are you 'animating' an existing one or working on it / tuning it in parallel with the legs?

So i’m using the 75054 AT-AT set. But i’m making adjustments to make it more accurate (more cockpit space for example). The goal is once i’ve finished the mechanism to isolate all the pieces that come with 75054 and then only use those for the body (i’ve made a more accurate version before just from said pieces).  All the while making sure weight distribution is good