Are Matt's Robot Hexapods Creepy or Cute? (Video)

Video Are Matt's Robot Hexapods Creepy or Cute? (Video) 35

Posted by Roblimo on Wednesday September 17, 2014 @06:03PM from the you-put-your-right-foot-in-you-put-your-right-foot-out-you-pick-up-a-human-and-shake-it-all-about dept.

Video no longer available.

University of Arizona grad student Matt Bunting doesn't come across as a mad scientist. That's a very good thing, because his robot hexapod creations are easy to imagine crawling across the USA in large hordes, devouring everything in their path and using all the electricity they come across to feed their Queen Hexapod, a 3-D printer mounted on a hexapod chassis that turns everything fed to it into more robots. Luckily, the real life Matt is an affable (self-described) "Roboticist, Electrical Engineer, Musician, and Rock Crawler" who freely admits that at this time his robotic creations have no practical application whatsoever. This is probably true, except for the fact that they can liven up a music video like mad, as you can see on YouTube in Pedals Music Video (featuring REAL robots) . Our little video is a lot simpler, of course. In it, we interview Matt and he tells us what he's up to with his robots, and gives some 'how to get started with robotics' advice for budding young engineers. (Alternate Video Link)

Tim: Matt, tell me a little bit about these robots that are sitting here to your left?

Matt: Yes, so I designed these in 2009 as part of a robotics class at the University of Arizona. And robotics class was geared towards implementing cognition in a machine. So the original class project was to have the Hexapod learn how to walk by using a camera. So ever since that original class project by – well, first I gave the Hexapod a makeover where I used a 3D printer and synthesized their legs in ABS plastic.

Tim: What were they made of before that?

Matt: Before I used a bunch of pieces of brass from Ace Hardware and I soldered them together, so it was highly imprecise. It worked, it got the job done for the class, but it looked a little crude.

Tim: Now the hardware that is running this, talk about the evolution that you have gone through for this?

Matt: So switching to the 3D printer ABS plastic definitely sped up a lot of time. I didn’t have to manually solder parts together, cut them anymore, I could just quickly hit a print them, but plastic was certainly a bit weaker than the brass pieces I was using. So I had to go through a few different iterations to figure out exactly where I needed to make it stronger. And I’m in fact still doing that. Here’s a small piece of plastic that just came off, and that’s just due to being a stress fracture, it had a form like this where it’s running for hours and hours at a time. So I'll have to go back and just add a little beef to certain areas. It’s certainly an iterative process. At first I don’t foresee certain issues, but the nice part is that I can make that change and then have the part the same day and test to make sure that it fits.

Tim: Now, how did you actually come up with – why did you choose the Hexapod? What are the advantages of Hexapod? Because you wanted to have a system for learning for science?

Matt: The advantage of a Hexapod dynamically speaking is that it’s very stable, it can always support itself versus say a humanoid which is very tricky in terms of balance. However, in terms of machine learning, that means I have six different legs I might need to learn for, so there is a lot of degrees of freedom that it has to learn for, so I don’t know if a Hexapod is necessarily the best platform for what I was doing, but I was able to get it to learn how to walk, so I think that’s the most important thing.

Tim: Now with all those degrees of freedom, I know you mentioned you had to put up a lot of motors in there? Talk about this.

Matt: So there’s 21 total motors; that means being six legs, there are three motors per leg, but then there’s a head mechanism and that’s where the last three motors are if you are trying to do the math, hey that doesn’t make sense. So each leg has three degrees of freedom, that means in this configuration I can define an XYZ location for each leg and as long as that’s within reach I can place it there. Mathematically I can figure out the exact angles that are needed to place it down. And similarly with the head, the three degrees of freedom are used to apply movement to the head mechanism so that it can always stay stable in case the body is twisted.

Tim: Now before we turn to the camera, we were talking a little bit about the components of this. It doesn’t sound like every hobbyist is going to make one of these. Can you talk about why?

Matt: So the costs add up very fast, each of these motors is $200 and there are 21 of them, so that’s immediately $4,200.

Tim: That’s a good strength.

Matt: That’s little bit, especially if it’s your personal budget, and so when I first started this I used cheaper motors but fortunately with Intel's support I was able to pick all the nicest motors.

Tim: What is the optical system?

Matt: So on the front the head mechanism, my main sensor, actually my only sensor is computer vision, so I have a machine vision camera, the camera is made by Point Grey, that’s a Chameleon model, and it’s very nice, has some advantages over a traditional USB webcam, it doesn’t use a frame buffer, and the images come out uncompressed. So the nice part is that I don’t have to use the computer to decompress the images and I won’t have like a delay of time of when the image was taken.

Tim: Let’s talk about the computer inside, you started out with an Atom-based system here?

Matt: Correct. So this is the original Atom-based computer made by CompuLab.

Tim: It's communicating via Bluetooth?

Matt: So, normally I use Wi-Fi to communicate; however I do have a Bluetooth dongle here, so that I can communicate to Bluetooth devices, such as a PlayStation 3 controller. But this is a full blown computer, it’s very nice, has a 6 USB and everything, and this is what I originally used in 2009; however, this has been replaced by Intel Edison entirely. So Intel Edison has a little Wi-Fi and Bluetooth chip onboard, also has its own FLASH drive and RAM, so we’re able to take this already pretty small form factor and make it micro size.

Tim: The purpose is pretty abstract it sounds like?

Matt: The purpose of – yeah, so this is actually... I'm going to grab this and flip it around because it’s pretty hot out here. Here we go, okay.

Tim: In first constructing it, is this pretty much an intellectual exercise?

Matt: Basically yeah, there’s no direct practical purpose for this, though I’m sure lot of people can find some.

Matt: Right, so the idea is to explore algorithms, specifically at the link between vision and brain as inspired from biology. So I’d see how biology uses vision and then I try to mimic similar behaviors for the Hexabot.

Tim: You mentioned that six might have been a high number to start with for this kind of project, are you also working with smaller numbers of plain robots?

Matt: I have in the past dealt with some bipods and quadipod, a cheetah robot specifically, but I still use Dmitri for a lot of the current projects, simply because it’s easy to not have to worry about balance. As long as you have three legs on the ground, you can do anything you want with the other three legs, and it will stay standing just fine.

Tim: It would be good at Halloween time?

Matt: It's perfect for Halloween.

Tim: You scare the kids with this?

Matt: Oh yeah, I put candy on the back and I make the kids just grab the candy, and before they grab it I make it jump, so.

Tim: If someone wanted to get into building robots, is there any advice you'd give them now knowing what you know?

Matt: Lego Mindstormsis certainly the best way to get into programming and using building blocks to assemble an electromechanical device. And after that you can get – if you want to use like a hot glue gun, you can certainly make things pretty easily using like standard hobby servomotors and make things walk around, so those are probably the best ways to get started, especially at a young age.

This discussion has been archived. No new comments can be posted.