Bruton elegantly bypasses computational complexity by embedding the robot's logic directly into its mechanical geometry. It is a masterclass in using hardware simplicity to solve what is traditionally a software-heavy problem.
Deep Dive
Voraussetzung
- Keine Daten verfügbar.
Nächste Schritte
- Keine Daten verfügbar.
Deep Dive
Anti-Inverse-Kinematics HexapodHinzugefügt:
This is the second in a series of designs that you can build. All of my projects are open source, but most people don't actually build things like the AT80 you can ride on. So, I'm doing a series of smaller, more accessible projects. The first one was a mini robot dog, which you can find in my channel, and now it's time for a hexopod, which is six legs. Last time I used an Arduino Uno, but a lot of people asked for a remote controlled version, so this time I'm using ESP32s.
I'm using standard RC style servos again which are easily available. All the cannon code is free and it's available on my GitHub. But I also have enhanced downloads for patrons which include useful things like a wiring diagram, a list of all the parts used including nuts, bolts, and screws and written documentation about the build and the code. But the designs are still open source. So there's nothing to stop you making kits and selling them or commercializing the design somehow else.
Which is totally fine. just like Rep 3D printers, the Linux operating system or Arduino clones. If you do want to do something commercial with it, then it would be nice if you could support me on Patreon if you wish, but you don't have to because it's still open source. There are just eight servos in this design, which is less than half the 18 actuators you'd normally find in a normal hexopod.
I'm publishing solid models for this project in step or STP format, which is actually what I do for every project.
These are far superior to STL files because you can load the project into your own CAD software and edit it as if it was your own project. So, it's really easy to pull the surfaces in and out or change it completely. Most hexopods like Matt Denton's Mantis walking machine have 18 actuators, which in this case are hydraulic cylinders. And each cylinder has two proportional hydraulic valves controlling it. So, there's a lot to control there. The control system has to sync the foot paths up with each other. So, we need a system of inverse kinematics. I've built several walking robot projects which used inverse kinematics like some of the larger robot dogs. This involves a calculation for each leg which allows us to place the feet where we want in cartesian coordinates and then calculate each joint angle for the knee, elbow, and hip to place the foot correctly. Then, to move the foot in a straight line, we need to interpolate through all the points on the way from one point on the path to the next and feed those positions into the inverse kinematics.
But what if we just built the robot? So all the axes are linear and move in straight lines to start with. Then we don't need to do any calculations. This hexopod uses racks and pinions so that the legs move up and down in straight lines and move from one step to the next in a straight line or a perfect arc.
Thanks to Polymaker for supporting the channel with 3D printing filaments. I'm using Polyite PLA here, but you could use almost anything. I've printed the racks laying on their sides, and that means that the teeth on them are going to be much better definition. I'm using bearings to make sure my racks run in a perfectly straight line. And these are pretty standard ones, which are 22 mm on the outside, 8 mm on the inside, and 7 mm thick. And these are basically the same as skateboard bearings, so you should be able to find them fairly easily. There's a little shim in the middle so I can put a selftapping screw in. And this is my rack. And of course, the bearings are going to run in the groove in the back of the rack, which stops the rack escaping.
I'm using the MG96 servos again, which are metal gear servos and seem okay. I got these ones from Amazon. They came with a plastic servo horn. So again, I've upgraded them to the 25 teeth aluminium ones, which are easy to get hold of. The servo screwed into a bracket. I've just used some selftapping screws for that. And there's a little hole to let the cable out. I've made this gear for it, which has a little recess for that aluminum servo horn. And it has two holes. One goes in the middle and one goes in the end of the servo horn. And that means we can put two socket cap bolts in and those will screw into the aluminium. One holds the actual horn on into the servo and the other one just goes into the lever. And that means that that gear will have lots of torque and it will never be able to slip.
Here's my working rack and pinion mechanism. And that seems to run okay.
And you can see those servos are pressed against the flat of the back of the rack. And the front of course is pressed teeth to teeth. And that means that that rack is actually fixed in there very well and it won't be able to slide out.
Also, that lever on the servo never hits the rack because the servo doesn't turn far enough anyway.
And obviously, if you need to change clearances or any tolerancing for your printer, then you can cuz I've published the solid models, which pretty much means you can edit it as if you drew it.
And of course, there's six of those, one for each leg.
These servos are apparently 12 kg cm servos, which means at 1 cm radius, if we tied a pulley round and hung a weight on it, we should be able to lift 12 kg absolute peak. So with about 2 cm, which is what we've got here on this linear arrangement, we should get about 6 kg of force. And that's actually quite a lot.
We shouldn't really be pushing servos to the limit. And we don't need anything like that because basically this is just a piece of plastic it's got to lift.
It's got three legs to hold it up. But that's more than enough and it should work well. Three of the legs have taller brackets and I've colored those in green. And that's cuz we have two overlapping layers again, the same as we did with the mini dog, which was the last video in the series. So, let's screw those brackets on and screw the legs on. And what we should find is that we have two tripods. And one of them is slightly taller. And that's cuz the taller one sits on top of the bottom one. And that makes up the whole hexopod with all of its six legs. As I said, this series of projects is designed to be accessible for as many people as possible to build because I believe in education and engaging with the community, but I've also done a number of other open- source projects which are much larger. This video is sponsored by Anyesk and they've sponsored a number of bigger projects on YouTube so far from other creators. I'd really like you to go to anyesk.com/science and submit a bigger project idea you'd like to see me build, which they'll pay for. They sponsored Styr Pyro's 100 car batteries wide in parallel, Jay Laser's self-healing iron man helmet, Chris Dole's disposable vape battery car, and Smarter Everyday's refueling a nuclear reactor video. So, imagine I had unlimited budget to build something no one else has done before. Check out my Star Wars AT80 that I can ride on. The bull bikes, the screw bike, or maybe my rhino tank that Colin Furs helped me test. Perhaps I could build another working Star Wars vehicle or robot or something else from science fiction. Go to anyesk.com/science and submit any idea that you'd like to see Anyesk sponsor me to build. And by the way, Anyesk is remote access software that allows you to access your computer remotely. And it works on Windows, Mac OS, Linux, including Raspberry Pi, Android, and Apple TV. But the project doesn't need to include any software that just really want to work with the community to build cool projects. As with the mini dog, the two layers slide against each other, of course, picking up its feet as it goes so that it can walk. And they also rotate against each other. So to facilitate that, we have a section with another piece screwed on the bottom. And I've done that so both parts could be printed flat on the bed because the top part has features on that slots into the slot. And you should find that that moves quite freely up and down.
You'll also notice there's a hole in that top part and a hole in the bottom layer. So, I've put an M4 bolt through with a lock nut, and that's going to make the pivot and the sliding section.
So, now the top and bottom layers slide up and down against the middle layer.
And they also rotate. That means we can turn on the spot and we can walk along.
And you'll notice that middle leg stays aligned with that partial spur gear, which is basically a curved rack as it slides up and down. And that means that we can go and put something on there driving it against the middle leg to facilitate the rotating motion. And of course, yes, that's going to be another servo which is going to be in a bracket running against that gear. And that fits in upside down just above the middle leg. And I've made a triangular section on it to match the profile of the leg. I put some uprightes in to hold that servo in the right place. And now the servo moves around that circular track, which is a partial spur gear. And I left little holes in the side so the lever on that gear, which is exactly the same as the others, can stick out as it rotates.
The linear motion is another rack and pinion. And that causes the top two layers to slide against each other. So that's how it's going to mainly take steps with the rotational axis for turning on the spot. So now we can move the legs against each other in a linear fashion or we can also rotate them. And that means we can do all of the motions apart from side stepping of course. So this looks like it's going to be quite a solid build with at least three legs on the ground at a time and all the axes already move in straight lines. So we don't need to solve any complicated inverse kinematics. We can just lift the foot and move it forwards and all the mechanics take care of only one motor being responsible for one axis. But it's time for some electronics. Last time I used an Arduino Uno and I built a simple programmer with buttons on the back of the robot. But lots of people asked for a remote controlled version. So this time I'm using ESP32s and ESP now to get the data from the remote to the robot which is very simple. ESP32s are very cheap. I found three packs on Amazon for less than £15 in money, so I'm going to use one in the robot and one in the remote. ESP now is a very simple way to get the data between them wirelessly. It's Wi-Fi, but you don't need to attach to any Wi-Fi base units. You can just send data straight to the hardware address of the other ESP32, and they connect straight away at powerup. I found a really good tutorial on this at random nerdutorials.com, which has example code, which works well. And I've left the copyright notice from these examples in my code for the wireless com section. First of all, we're going to use some code to get the MAC address of the receiving ESP32, which we run on that receiving ESP32.
And then that address goes into the transmitter code, so it sends the data directly to the correct receiver. I've wired two axes of an analog thumb stick to the analog ins of the transmitting ESP32, and I'm powering the VIN of that ESP32 from four AAA batteries. This sends a data structure to the receiver where I'm writing to the serial terminal so we can see the data. I've built this project on perma protoboard which is the same layout as breadboard and there are various types available. Unfortunately, it's not quite wide enough for there to be spare pads on both sides at the top cuz the ESP32 is too wide and that means some of the wires need to be attached on the bottom of the board. I'm using a 2S lipo to power everything here. There's a switch that turns the battery power onto the VIN of the ESP32 and also the VIN of a power regulator which is a turning S-spec and that's going to power all the servos along this power rail. I've used pieces of pin strip heat shrink and wires to make connectors for my servos. And in fact, I've attached those to digital pins on the ESP32 on the breadboard where pads are showing on the top on that side. And that means I didn't have to put any wires on the back of the board after all. The code for this is pretty simple. I've got one multitasking state machine which repeatedly runs through the loop with no blocking functions like delays or while loops. This checks the system clock as well as which state it's in so that the walking sequence picks up the legs and moves them in the right order at the right time. There's a section which moves the linear slider or the rotation axis of the robot depending on the joystick value. So, we don't need different sequences for walking straight or turning. I wrote this code in a similar way to the Adafruit Arduino multitasking tutorial. So, if you check that out, it'll be easier to understand what's going on. I also scaled the values for how fast it goes and how far the axes move based on the joystick motion. So, if you walk slower, it picks up its feet more. Speaking of feet, I made some feet out of Polymaker Polylex TPU90. So, it should have more grip than the smooth plastic. So, now we can see it walks pretty well. We've got various speed modes, and this is pretty much full speed with the shallowest step, but I'm pretty happy of how that works.
It can turn on the spot pretty well. And I'm really happy with that turning action. It can't walk in a curve, unfortunately, because if we rotate that axis and slide the legs, then the legs hit each other. But I'm pretty happy of how far it turns to do a 180.
And we can see that rack and pinion sliding axis working pretty well. So, we get quite a big stride length, which is about 60 mm.
Really happy of how that turns. I wasn't sure how much of an angle we get in there before the legs hit each other, but we're getting about 25° in either direction. So, we get a 50° turn in one step cycle.
If you push the joystick less in any direction, then it goes slower and it picks its feet up more. So, that means it can step over obstacles quite well. I guess if we had bigger spurgears on our rack and pinions, then we'd be able to pick the feet up even further and have even longer legs, but that's pretty good for now. And you can see it's stepping over all the lumps and bumps in the grass.
Tripods are of course always stable and we've got at least 3 ft on the ground at a time. So, it's a bit like a stool on an uneven surface, which means it always settles and molds itself to the terrain as it goes. So, that makes it pretty stable. And if we pick up the feet a lot, we can walk on almost any terrain.
Ähnliche Videos
Beyond Robotics | European Rover Challenge 2026
beyondrobotics
189 views•2026-06-01
Beatbot Sora70: JetPulse Technology and AI obstacle avoidance and navigation!
DroidModderX
26K views•2026-06-02
Tesla FSD 14.3.3 Hits Phoenix Streets - FIRST LOOK
anthonystesla
114 views•2026-05-29
NVIDIA Just Unveiled a 75 DOF Humanoid Robot
DPCcars
74K views•2026-06-01
Elon Musk Just Revealed Fremont Line for Optimus Gen 3 Mass Production
TheAINexusOfficial
180 views•2026-05-30
人機一体「零式人機 ver.2」 子ども企画【おもしろ発見!モビリティー】 #乗り物 #automobile #robot #shorts
KyodoNews
1K views•2026-05-28
China’s New Luna AI Robot Looks Shockingly Human...
NextGenHumanoids
850 views•2026-05-28
Reachy Mini: the $300 open source robot you can actually hack — Andres Marafioti, Hugging Face
aiDotEngineer
662 views•2026-05-29











