It really feels like there's not enough hours in the day for the amount of planning, printing, building, coding, and testing you do week-to-week for these videos. It's honestly incredible and also a bit scary. I think you deserve a nice month-long vacation, friend. Really great work!
I highly doubt he prints all those parts himself. Every once in a while we get a quick shot of one printer but i doubt that's the whole story. But his whole shtick is this DIY veneer so whatever....
you can see in other videos that he has a rack of 6 or more lulzbot printers. It is indeed a lot to print but his sponsor seems to have provided enough 3d printers for these projects. Incredible amount of work though, really quite jealous of his discipline.@@sprinteroptions9490
I think if you had six, so that every 3rd section is in sync, then you'd get the best results, as there's almost constant forward motion and always two sections lifting up the rest of them
James, in order to make a worm that you can ride on you'll need to keep in mind the issue of mass. Segments with greater mass, i.e. where you're located, will be more resistant to change when contracting. You may be able to solve this, and potentially improve the locomotion, by designing an "upper deck" where you sit. By syncing your placement over the segment that is expanding at the time but shifting forward for contraction you would encourage forward movement.
When you went from step input to ramped input, you needed to add a delay in the routine for each movement to finish before the next movement starts. Because of this, you essentially always have all your parts moving at once. No section of the worm ever has a stable platform to exert force against. If you add in the delay, you can trigger the segments sequentially, with each acting against a pair of stationary partners. The mechanism can take much larger and more stable steps.
If the wheels had grip and were designed to rotate in only one direction. Then they would free wheel going back and then drag against the floor when going forward. That should really help with efficiency.
Have you made a centipede yet? Because that could be programmed to do peristalsis. They're also incredible to look at. Whenever i find one i spend a little time to observe the perfection of that movement, it's really incredible how nature bio-engineered such a creature.
I was literally thinking this exact same thing when he mentioned the feet and then was like nahhhhh, I'm like sir, the centipede is obviously the divergent evolution of the worm!!! Cool they can also peristalse as well.
Can we just take a moment and appreciate how good James is looking recently. Have you been working out mate? Whatever you’re doing…….keep doing it! Well done!
You made a complicated new mechanism to stop the feet sliding and It would have worked. And then you put wheels on it... you are a robotics artist. Love your content, it's fascinating.
I love this, so simple and very effective. I can see imagine with a little further development, this could be your future lollypop person, helping people safely cross a road. Brilliant keep up the great inspiring work. 👍
Could you make a strandbeest version? I think it would be a good compromise, and get more motion over time. It would be fun to still use the linkages type for steering too!
Would have liked to see a 750ms delay so that it would be a full even cycle between front and back movements. Exactly half of the 1.5 second one. That way there should always be one actively pushing forwards. Another idea would be to have the middle two segments have TPU feet so they have more traction, while the outer two keep lower friction feet since they aren’t supported.
if you made the very first and last roller move but with a freewheel mechanism then as it pushes the droopy front section along the roller has no friction and when it is subject to drag the freewheel locks which would give it grip - same for the very last roller so it gets real good one way motion - if u wanted to get fancy u could have these done like the gears on a ratchet so that you could activate them via a servo so you could get forward and reverse motion
It seems like the worm is moving all it's front parts forward sequentially, then pulling the back end forward using the combined friction of each forward-facing section. Do something similar, where you're effectively making a sine wave that terminates flat on the stopping part, allowing each back-facing portion of the vehicle lift up and move forward with no friction, while the front portion holds the rest of the vehicle in place.
Love the idea of the feet at 9:48 - It’s kind of unsatisfying right now since it kind of slides around. I think the feet would look even cooler, especially since they’re so unique
May I suggest replacing the rollers with brushes with angled bristles? Rollers move more or less forwards and back with equal ease. Don't you want something where the forces on the ground have a directional preference?
Brilliant idea, but perhaps have them expand and contract in sequential order with overlapping cycles, 1234 (All contracted) 123 4 (4 to 25% extension) 12 3 4 (3 to 25%, 4 to 50%) 1 2 3 4 (2 to 25%, 3 to 50%, 4 to 75%) But I think the ends will need a brake system of some kind to anchor the end going forward and grappling the rest along behind. For maximum efficiency, I suggest a reach / anchor / reach approach.
That last design you showed in the software with 2 feet on two rails looked like a great starting point... BUT instead there are 2 feet on each rail that make up each of the 4 segments that pivot with the ball and socket like in the peristaltic prototype. And each of the feet can lift up and down just enough for the 4 feet to be used in a 'walking pattern' then each segment is angled left or right to steer..... and you may even use 2 of these, each like the track of a tank with you sitting in between..... - I don't think you need any type of joints like a knee or ankle if you're on a hard flat surface, but if you use it 'off-road' you might be able to get away with a sort of 'shock absorber' in each of the feet to walk on uneven terrain, so that each foot's lift/extension doesn't require full extension to 100% ... but that would require position sensors on each foot.... but this is all obvious to you LOL - I know nothing about this, except for what I've learned from watching your videos..... - - I only wrote this out because you mentioned how it would look while moving, and I think this would look hypnotizing the way millipedes looks... or you could think of each section as a tiny horse lol (how that for a decoration - lol - you riding atop 4 tiny horses ) - ANYWAY ... I hope this at least gave you some inspiration or ideas... that's the least I can try to do for all that you have given me with your amazing videos and BRILLIANT projects!!! THANK YOU!
This works similarly to a simulation I made for peristaltic motion. It was for a videogame where you'd play as a loose length of intestine, but that's beside the point. The point is - I was using 6 sections for mine. I spent AGES trying to get it to do more than just vaguely wriggle around. What I found was that an even number of sections is just... bad, somehow? Three is useless, so 5 or 7 are a good pick. Slap another module onto the back of that worm of yours and it might well perform a bit better... Edit: I also remembered as I was watching the end that yes, offset = cycleLength*0.5 is the only workable one I found - until you have so many sections that offset*sections > cycleLength*9 or thereabouts, but then the simulation ebcomes unwieldy to use practically. Or, in your case, you run out of plastic and servos.
Nice to see the progress of an earthworm into Minecraft's lookalike digitalised reality. Some less sturdy connection between the sections could have brought the rolling bars back to earth😉 Would love to see some generative design in your printed products to use less filament than rectangular blocks. Would give it a more calculated, pre-thought form 😉
I like the idea at the end with the "feet", but maybe it can be more leg-like with a knee instead of ankle?? I think it would really cool with a line of legs trampling forwards. Another suggestion would be to use 3 different phases instead of just 2, since it's hard to define a movement direction with just 2 phases. ♥
you should look at a worm for refrence. it works in sections. you should try first two at like a 500ms delay, and then the next two at like a 1000ms second delay from that, but with a 500ms delay from eachother
Idea: just two (grippy) rollers per section. Both are on their axles with one way bearings. In opposite directions. Uppon moving towards each other the front roller cannot roll backwards so it pulls the section forward. On extension the back wheel cannot roll backwards so it pushes the section forwards. Wouldn’t this make an extremely simple solution for peristaltic movement? You could keep three wheels if really needed by making the central roller free to rotate in both directions.
This would benefit from using gears rather than servos. Convert rotational motion into linear motion like a piston and use that to make your mechanisms move.
Move the stem all the way forward, then move the first segment, when that is done going forward, move the next and so on, that way you are only moving one segment at a time while the rest supports the move. Or that's what my mind keep telling even if I really don't have a clue :D
I'd be interested in seeing a version without the concertina mechanisms and has ratcheting wheels. I think if you got the timing right each section could push in sequence and it would really move.
Perhaps if the round sections on the bottom had slight rachet gearing; it might easily slide one way but encounter more resistance the other. Not sure how you'd do it but hey.
The worms actually have a part of their body expanding, and then at the same time contract. Also parts of a couple of nodes that are expanded at the same time. Maybe you can have the walking animation only use a part of their lateral movement, and have the 'extra' room to work in groups
Would it help to put a breaking system on the rear wheels? It’s just like walking, the back leg pushes the body forward, by adding a breaking system (wheel locking system) you can keep the wheel stationary for a moment in time.
I like the foot lifting idea. I wonder how it compares for power efficiency with the peristalsis imitating version. Is one more efficient or would they be about the same?
Feels like the concertina bits are hindering the overall movement, and removing two of the yellow cylinders so they are more leg like would actually improve the locomotion
This doesn’t seem to actually be moving by peristaltic motion. It looks like the concertina sections are just a mechanism to lift each leg (or two at a time), move it forward and then set it down, so it’s actually just walking. Am I wrong?
You have a motion system which operates in three phases. You need three worker groups, each group on a separate phase, each member of the group executing its phse in parallel. Group A B and C, with ideally 2+ members per group.
Instead of a second motor, you could add something to lock the moving parts to one end or the other. It might become lighter and more energy efficient.
Could you use sprag clutches to lock the rollers in one direction? You're rollers look like the problem here. Like you said, the worms have one way hairs to help them move forwards.
This made me think: What if you were to make rideable robots that took the form of fantasy creatures that could be solar powered? Imagine being able to make them light, but sturdy, enough to carry folks around town using only solar energy. I mean I personally would love one lol
If you wanted, you could have changed the beams to be right angles, resulting in extended and contracted length being the same length. Henry Segerman did a video on something similar to what I am describing.
As long as you don't mind it only going in one direction, I think asymmetric gripper feet would work better. I.e. the front being a smooth curve giving low friction. And the back being vertical with a sharp angle. Obviously this would need a surface with some give to it. So carpet or many types of outdoor surfaces.
What about airbags? Rubber pads on the bottom that can inflate and deflate, so when they're trying to grip the ground they puff up. But when they're trying to be dragged freely, they deflate. If you did a flow through system where they're fed by an air compressor and then let the air out when they deflate that might be too noisy, so instead you could do a big air cylinder like a syringe that would be actuated by a motor, so each bag would have its air circulate back and forth between the bag and an accompanying air cylinder on each segment.
I think your original design is the way to go, where the segments get longer as they get thinner and lift off the floor. Except you’d need many, many more to make it work fluidly. There should be at least two ot three contact points with the ground at any one time, and the mass should translate from one contact point to the one ahead like in a slinky.
With Steve Mould's recent video about self assembling materials I wonder if something like that could be used for propulsion. Expanding up/down to make a lateral peristalsis or something?
I think the original way of doing it is more visually interesting, where the movement is done in pairs of sections, keeping the total width between the two sections the same, and having one thin section at a time moving backwards along its body, rather than doing it in pairs with a walking kind of motion. Perhaps with some kind of structure to keep the joints level with each other. I think it would be an interesting challenge to design such a structure than still allows it to turn.
Maybe a snake would be another yet similar form of locomotion to examine for this kind of build? With something along the belly of the robot/vehicles moving instead of the whole body acting like an accordion? Probably with slots in the belly, and fins or such that move in an offset circular motion so they come down and pull/push forward or backward?
I don't know why but the 500ms delay seemed a bit more centipede like to me than the 1500ms delay. Not to say it's more effective, maybe more organic is the word? I get the feeling this sort of thing gets more effective wit more segments, so the effect of the ends is minimized.
Also that is a FT of filament. 3DP isn't _always_ the best solution for making parts. Especially when what you really needed was metal and not plastic because stremgth to weight ratio is a thing.
Was watching this and wondering if the wheels could be made is such a ways that when they were going in the pulled direction they would roll and when they go in the push direction they don't roll.
A couple ideas to consider number one more sections this should help with the stress on the front and back. Observe how it is done in nature more detailed that way you can get the timing right
Maybe implement some telemetry and you could do some reinforcement learning to optimize timing? Then use that understanding to make the full-sized version as good as it can be
Couldn't you have tried 3 seconds? So the first one moves, then the second one, and so on. it would be slower but there would always be 3 legs supporting so a lot more consistent.
The last idea, two feet in parallel, is a step closer to just making the linear slider equivalent of a strandbeest. It's not a bad idea but isn't anything all that interesting. I'm not even sure if a paralleled peristaltic would still be considered a peristaltic contraption as now you'd be twisting that definition to fit a multi-pedal machine.
Instead of mechanical segments, why not consider something like inflatable bags? It would require valves to allow each bag to inflate/deflate to mimic the peristaltic motion.
With the lifting motion, the expansion along the length seems to be mostly irrelevant. It reminds me of the walking motion of those old cheap robot toys, just placed along a worm.
What are you doing James? For every action there is an equal but opposite reaction. You need to think friction. Probably a wave is hidden in the sync. Just go back to basics.
