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

[oracid]

15 hours ago, Nelson said:

I had a look at your walking models. Do any walk one leg at a time?

Unfortunately not. My quadrupeds use a gait based on trotting. This gait is based on a 180° phase shift between the front and rear legs, and left right. This is the easiest gait to implement with servos and independent legs.

In your case, I would try to study the gait of walking with an ultra simplified mechanism with a single and small motor that you could then apply to your AT-AT. The video posted by @Davidz90  is an excellent basis for reflection, even if it is based on a Theo Jansen mechanism which is quite complex.

In my opinion, the enemy is weight and the secret to success is to start small.

[LegoTT]

On 10/30/2024 at 10:12 PM, Nelson said:

you'll see some gears in there that are only a quarter of 360 degrees. Unfortunately, there is no Lego analogue of this.

Would this help ?

 

[LegoTT]

There is also this yellow contraction (only two yellow gears and one red knob needed) that is way more compact and can handle a good torque.

No solid gear immobilisation for the 3/4 non-engaged, though.

Edited November 14, 2024 by LegoTT

[Nelson]

@LegoTT These are both really interesting ideas. I will ponder them. Thanks for some genuinely good stuff!

Edited November 15, 2024 by Nelson

[Nelson]

@BrickBear Have you done any more work on your design? I think I'm going to try to get a working version of my cam/lever design going even if it's not capable of moving the final aesthetic model. I just want to to prove to myself that the concept is viable at some scale/weight. I had an idea the other day that may be pertinent to both our designs. I think that one reason my version with the treaded cams "worked" (I use that description loosely) was because the cam contour was slightly off in a desirable way, just a bit out of round/flatter in a few key areas. I'm thinking that if I can get a version where the leg on the opposite corner of the leg lifting also gets the weight taken off it, it will balance much like a 2-and-2 leg version walker would. For example: when the left front leg lifts and moves forward, the rear right leg raises (not moving forward) just enough not to carry any real weight, but not enough to come out of contact with the ground. This would leave all the weight on the right front and left rear leg, balancing the walker, just like a 2-and-2 walking design. I think I can achieve this with a cam design that has two flat spots rather than just one. In my current version, the large flat area of the cam allows the leg to be lifted by a lever during it's step. But there could also be a much subtler/shallower flat area that acts to lessen the downward pressure during part of the cycle while the leg is on the ground and the opposite corner is lifting. Not sure if I'm explaining this very well. I'll upload some photos and video once I get some new parts in. Lego has release several parts in the past year that are allowing me to make more eccentric cam designs.

Edited November 17, 2024 by Nelson

[BrickBear]

16 hours ago, Nelson said:

@BrickBear Have you done any more work on your design? I think I'm going to try to get a working version of my cam/lever design going even if it's not capable of moving the final aesthetic model. I just want to to prove to myself that the concept is viable at some scale/weight. I had an idea the other day that may be pertinent to both our designs. I think that one reason my version with the treaded cams "worked" (I use that description loosely) was because the cam contour was slightly off in a desirable way, just a bit out of round/flatter in a few key areas. I'm thinking that if I can get a version where the leg on the opposite corner of the leg lifting also gets the weight taken off it, it will balance much like a 2-and-2 leg version walker would. For example: when the left front leg lifts and moves forward, the rear right leg raises (not moving forward) just enough not to carry any real weight, but not enough to come out of contact with the ground. This would leave all the weight on the right front and left rear leg, balancing the walker, just like a 2-and-2 walking design. I think I can achieve this with a cam design that has two flat spots rather than just one. In my current version, the large flat area of the cam allows the leg to be lifted by a lever during it's step. But there could also be a much subtler/shallower flat area that acts to lessen the downward pressure during part of the cycle while the leg is on the ground and the opposite corner is lifting. Not sure if I'm explaining this very well. I'll upload some photos and video once I get some new parts in. Lego has release several parts in the past year that are allowing me to make more eccentric cam designs.

Unfortunately haven’t got round to anything yet apart from some mental visualisation of necessary changes and bricklink scouring. We are having rooms painted this week so need to keep things tidy and stuff has to be moved about to accommodate.
 

But I was looking at my model today and realised that while my front right leg is raised that the right back leg begins raising too at a certain point. So i’ve got a little configuration necessary to counteract that, speings I think in the leg where the feet connect, not enough to cancel out the feet being lifted clear off of the ground but enough to prevent unwated liftage.

[Nelson]

@BrickBear Do you have any video of your whole model in action with all four legs going? I went back over this thread, made a reconstruction of parts of your leg design, and spent the evening studying it. I think we could combine some portions of your mechanism with mine. We are both using a rotating lever against a swinging lift-arm to raise the leg. The main difference is that you have a piece of linkage moving the leg back and forth while I am attempting to use the lifting lever to move the leg forward laterally in addition to raising it, and then using a slow-moving gear to move it back. The forward lateral movement part of my design is where I'm having challenges, so I'm going to try and incorporate your linkage idea instead. I think my cam design is still applicable in keeping the leg at the desired elevation while in its down position. I've also used springs to do this in a similar way to your model, and had success, but found the weight was too much after adding the bodywork. It may take me some time to figure out how to combine the two. I have the hinging "hip" point of the leg below the levers and linkage while you have it above. I have a "stay-cation" coming up and can dedicate some real time to this.

 

[BrickBear]

@Nelson

 

Unfortunately I don’t have a video of it yet due to balance. I could try powering it while holding it up just above a surface so that you can see it moving. Initially I did try to use cams keep the legs down, however it was before I had the linkage connection right so it seemed to not be stable enough and the timings made it not as smooth as I liked.
 

However now I realise the linkage position had to be a certain way. if you can work out a way to do this successfully timed I’d be very interested as it may turn out to be more successful than initially thought.

 

Edited November 18, 2024 by BrickBear
edit

[Nelson]

@BrickBear I've been working up a  few CAD models that incorporate features of both our designs. I had a thought last night and did a bit of research on gyroscopes. I wonder if adding a small flywheel inside the body that pitches forward and back in sync with the left and right leg motions could help with balance. It would really add some complexity and there probably isn't enough room, but it was an idea.

[BrickBear]

9 hours ago, Nelson said:

@BrickBear I've been working up a  few CAD models that incorporate features of both our designs. I had a thought last night and did a bit of research on gyroscopes. I wonder if adding a small flywheel inside the body that pitches forward and back in sync with the left and right leg motions could help with balance. It would really add some complexity and there probably isn't enough room, but it was an idea.

In a similar vein I was considering adding counterbalances that swing left and right. Basically you’d have the hips going up and down and then you’d have each hip attached to a see saw like mechanism and when one leg lifts up it would seesaw the weight away from it to the other leg.

So left leg goes up, seesaw left side goes up, weight leans right.

 

 

I also had another idea which was to make the legs move forward like a skateboard. Basically when the skateboard platform leans to the right, the front left part of the axle moves forward as if to steer to the right (but you could modify it so the leg stays straight) and as the body would be leaning right it would act as a counterbalance to the left leg moving forward.

[Nelson]

@BrickBear The "skateboard" idea is interesting. I'm curious to know how you'd pull that off mechanically. I've experimented with using the battery box as a moving counterweight, but there's just not enough space within my model's body to move it around much and, ultimately, it can't be moved to a position far enough over to really do much good. For me to realistically try a gyroscope, I'd have to rethink my motor layout to make it easier to get power to the upper area of the chassis. I should have some new designs put together in the next few days. Hopefully, the project will move forward a little more.

[BrickBear]

1 hour ago, Nelson said:

@BrickBear The "skateboard" idea is interesting. I'm curious to know how you'd pull that off mechanically. I've experimented with using the battery box as a moving counterweight, but there's just not enough space within my model's body to move it around much and, ultimately, it can't be moved to a position far enough over to really do much good. For me to realistically try a gyroscope, I'd have to rethink my motor layout to make it easier to get power to the upper area of the chassis. I should have some new designs put together in the next few days. Hopefully, the project will move forward a little more.

Sounds good 👍

As for the skateboard mechanism…well maybe something like this?

Edited November 22, 2024 by BrickBear

[Nelson]

This version uses a crank/piston-style lever inspired by the @BrickBear designs (the red connector with the black axle) to move the leg forward quickly, then back slowly, rather than trying to get the lifting arm to accomplish that lateral motion as in my previous models. This version also incorporates a new cam design that releases the downward pressure on the leg opposite the one lifting. This means the model is always balancing on two legs rather than three (even though three are still in contact with the ground). This helps to level out the overall balance and prevent the walker from leaning toward the raised leg. These ideas show promise, but are not without issue. The crank/piston lever needs to move the leg further than this to actually propel the walker at a reasonable rate. The lifting mechanism is (still) underpowered and jerky. Perhaps it's time to get a couple XL motors into the design and maybe try a piston-type lifting action rather than this lift-arm lever design. Also, when I attach the real legs, they're too heavy for the mechanism to move effectively. Still, this feels like progress.

 

Edited November 26, 2024 by Nelson

[BrickBear]

On 11/26/2024 at 1:02 AM, Nelson said:

This version uses a crank/piston-style lever inspired by the @BrickBear designs (the red connector with the black axle) to move the leg forward quickly, then back slowly, rather than trying to get the lifting arm to accomplish that lateral motion as in my previous models. This version also incorporates a new cam design that releases the downward pressure on the leg opposite the one lifting. This means the model is always balancing on two legs rather than three (even though three are still in contact with the ground). This helps to level out the overall balance and prevent the walker from leaning toward the raised leg. These ideas show promise, but are not without issue. The crank/piston lever needs to move the leg further than this to actually propel the walker at a reasonable rate. The lifting mechanism is (still) underpowered and jerky. Perhaps it's time to get a couple XL motors into the design and maybe try a piston-type lifting action rather than this lift-arm lever design. Also, when I attach the real legs, they're too heavy for the mechanism to move effectively. Still, this feels like progress.

 

 

Looks promising. Certainly try an xl motor. 

Do you think a medium + hard spring might be useful for push down?

There’s also the question of linear actuators. They can do some interesting stuff: (note video not mine)

 

Edited November 27, 2024 by BrickBear

[Nelson]

Well. I worked up a few more designs, one using Power Functions XL motors, one using large Powered Up motors. I experimented with a piston-style lifting action rather than the liftarm method I've been using. Nothing improved the functionality. The XL motors were overkill. Two large PF or Powered Up motors are more than adequate using worm gears. The piston style lifting motion still had the same issue as the liftarm where it flips though the lift cycle too quickly because of flex in the drivetrain. The lifting action gets backed up then jumps through to the end of the lift rather than creating a nice smooth lift and set-down. The one new bit of info I discovered is that the cams don't need to hold the leg in the down position for 3/4 of the cycle. They really only need to hold downward pressure for 1/2 the cycle. The key is to have the legs across and behind the lifting leg supporting the model while the one on the opposite corner is unsupported. This makes it so the walker is only supported by two legs at a time and the one on the opposite corner of the one lifting isn't inadvertently pushing the model over toward the lifting leg.

Here is another design I'm going to try. I'm calling it the "slide-piston" design. It also will probably not work.

 

[BrickBear]

On 12/11/2024 at 10:21 PM, Nelson said:

Well. I worked up a few more designs, one using Power Functions XL motors, one using large Powered Up motors. I experimented with a piston-style lifting action rather than the liftarm method I've been using. Nothing improved the functionality. The XL motors were overkill. Two large PF or Powered Up motors are more than adequate using worm gears. The piston style lifting motion still had the same issue as the liftarm where it flips though the lift cycle too quickly because of flex in the drivetrain. The lifting action gets backed up then jumps through to the end of the lift rather than creating a nice smooth lift and set-down. The one new bit of info I discovered is that the cams don't need to hold the leg in the down position for 3/4 of the cycle. They really only need to hold downward pressure for 1/2 the cycle. The key is to have the legs across and behind the lifting leg supporting the model while the one on the opposite corner is unsupported. This makes it so the walker is only supported by two legs at a time and the one on the opposite corner of the one lifting isn't inadvertently pushing the model over toward the lifting leg.

Here is another design I'm going to try. I'm calling it the "slide-piston" design. It also will probably not work.

 

You’re doing gods work. Meanwhile i’ve finished my course and the house decorating is done so I should have a little time for development at the weekend.  I’ll consider your discovery on the leg lifting situation when working on it. 

[Nelson]

3 hours ago, BrickBear said:

You’re doing gods work. Meanwhile i’ve finished my course and the house decorating is done so I should have a little time for development at the weekend.  I’ll consider your discovery on the leg lifting situation when working on it. 

[BrickBear]

So this is what I worked on today:

 

 

It’s essentially a redesign of my walker iteration’s chassis, reinforced and the suspension repositioned to take a heavier weight. The bend in the middle is also more stable than it was before. Few bits to finish adding and a tweak or two. But this should provide me with a better foundation going forward and has refreshed my mind on the project after being away from it.

 

The previous one for comparison: 

Edited December 14, 2024 by BrickBear

[Divitis]

23 hours ago, BrickBear said:

So this is what I worked on today

Can we see it moving?

Pretty pleaseeeeeee

[BrickBear]

54 minutes ago, Divitis said:

Can we see it moving?

Pretty pleaseeeeeee

It’s not quite finished yet, I have to adjust some stuff for it to move but once done i’ll make a gif. 

[The_Cook]

Is there a reason that you're using black friction pins as the pivots for the beams that the shocks are attached to? If you're using black friction pins for pivots then that's going to make it a lot harder to drive the mechanisms.

Of course there may be valid reasons that I don't immediately see why the friction is important, feel free to educate me if that's the case ;-)

[Aurorasaurus]

1 hour ago, The_Cook said:

Of course there may be valid reasons that I don't immediately see why the friction is important, feel free to educate me if that's the case ;-) 

I'm not sure if what I'm about to say applies to the AT-AT mechanism, but when building fast rc stuff friction pins help reduce rebound in suspension and keep everything moving at a slightly more regulated speed. Could be something to do with that.

Another reason could be for inserting a bar, and extending the pin, which cant be done with the grey ones.

[BrickBear]

1 hour ago, Aurorasaurus said:

I'm not sure if what I'm about to say applies to the AT-AT mechanism, but when building fast rc stuff friction pins help reduce rebound in suspension and keep everything moving at a slightly more regulated speed. Could be something to do with that.

Another reason could be for inserting a bar, and extending the pin, which cant be done with the grey ones.

The real reason is much simpler, I have many black pins and fewer grey ones in my main technic boxes so I like to use them for things where it’s absolutely necessary for the frictionless stuff. 

[Nelson]

I incorporated the slide piston design I posted earlier into a full chassis and it actually worked fairly well. There may be some promise in the concept after all. Unfortunately, it only walked through about 5 full cycles before a tooth broke off the main gear transferring power to the leg mechanisms. It's a real pain to change it out, pretty much a full disassembly. I'll try to get to it tonight and get a video up soon.

AT-AT Slide Piston Design by Nelson murray, on Flickr

AT-AT Broken Gear by Nelson murray, on Flickr

[BrickBear]

Ok so I’ve nearly finished modifying the frame. I’ve added the gearbox mechanism to adjust the front and back cams independently of each other, i’ve added an L-Motor and repositioned the grey technic frames one hole over as stated last time. 
 

one problem i’ve run into is the l motor isn’t really powerful enough with the resistance of the springs having increased. So i’m going to upgrade to the xl motor. Otherwise, the mechanism is now stronger and more polished than before and the positions of everything should be enough to support a battery box, steering motor and ir reciever as well as bodywork. 

2 minutes ago, Nelson said:

I incorporated the slide piston design I posted earlier into a full chassis and it actually worked fairly well. There may be some promise in the concept after all. Unfortunately, it only walked through about 5 full cycles before a tooth broke off the main gear transferring power to the leg mechanisms. It's a real pain to change it out, pretty much a full disassembly. I'll try to get to it tonight and get a video up soon.

AT-AT Slide Piston Design by Nelson murray, on Flickr

AT-AT Broken Gear by Nelson murray, on Flickr

That is very impressive. I’m guessing you’ll have to upgrade to a thicker gear if possible?