[youtube=http://www.youtube.com/watch?v=QBYT2PeLxUk]
We love it when footage of a robot prompts a “holy crap” response from us. This little guy, a product of the Chiba Institute of Technology, uses four rods as a suspension system for jumping. The bulk of the bot can be moved up or down, using its momentum to raise the wheels and jump to the next level. Check out the clip after the break to see how getting down involves a controlled fall as graceful as a dancer. Doctor Light better get cracking on another robot to take this one out when it turns on us.
[youtube=http://www.youtube.com/watch?v=Qf4kc95rCbc]
Wow, that is insanely cool. I was not expecting motion that smooth.
I want to say this is impractical, but its really gonna be the basis for future development of robotic balance controls.
They should have put a Mario custom on it and add 1up Mushrooms ;)
All the cool shit comes from Chiba City.
Holy crap! What a kick ass idea! I’m sure that will fuel a TON of ideas to the robot builders that visit this site!
does anyone know that hack group in toronto.
sorry to hijack comments
ty
Awesome robot, awesome Dr. Wily reference.
I feel like their control system isn’t particularly stable. The settling time after the jump was too long; with some tweaks I bet they could reduce some of that bounce. That being said, what a great idea!
I want to see them put it on a skateboard and have it do that.
Now if I can only get this system to function in my black Pontiac Trans-Am…. We’ll have you back soon, Kitt…
That is impressive, and prolly the first of it’s kind!
any more info? does anyone know what the actuators are?
lmao, i love it. and it definitely makes me wanna grab a p-buster and shoot it
wild, and if I had my guess, springs in the tubes
@Brennan
I think that is by design. Look at the first video — it took multiple oscillations to reach the desired height. The important thing to look at is how stable the robot is once it has reached the next platform.
http://www.youtube.com/watch?v=DwQbPgouUYo
@ Bob, re Trans-Am
Look into the Bose powered suspension system, it can do this on a car, real video here:
http://www.youtube.com/watch?v=eSi6J-QK1lw#t=1m42s
Sorry the time tag didn’t work, go to 1:42 to see the car hop an obstacle.
@nubie
Bose, the marketing company that also makes speakers, now makes suspensions?
Unless they call this robot JumpMan, Wily had nothing to do with it. (Or maybe Wily finally figured out that calling your robot what it primarily does is the worst way to keep it from getting countered really damn easily… Noone would assume something called Snakeman would use a flamethrower)
@NatureTM and Mitsubishi makes computer monitors.
Also, what I want to know, does the robot use the markings on the floor to know when to jump, or does it calculate using cameras?
http://www.youtube.com/watch?v=pwgOG160Vfk
That really is incredible. Definitely a “holy shit” moment!
OMG Double jump FTW! Always thought it was just an invention by game designers, but it seems it can really be done. Awe.some robot
i think all of the human race looks like goombas to it.
Video says ‘copyright 2008’
Not only that but to me there’s looks to be something fishy about it, the movement looks like it is stabilized by something unseen.
@Whatnot that’s probably just an illusion (sortof) since most of the video is slow motion (plus if you look more closely, it does wobble a bit). aside from that, it’s just extremely balanced so the jump doesn’t exert any torque. you probably all noticed this before i did, but it looks like all it does is pull down the mass with a string and then release it on springs (Osgeld), so it’s not very flexible. i mean… you could wind it up to different degrees and stuff, but like Brennan said, the settling time isn’t optimal and the reset time is kindof horrible too. given how “proof of concept”-y this seems… it’s probably just jumping based on the lines, since they’re testing a jumping car, not image processing. that, and… the lines are there.
still, this is cool to watch. i wonder what all of the hardware on it is for, since it seems to have a lot more computing power than a trigger-and-release system should need… it does make you realize that we could be doing a lot better at robot motions. if this could be accomplished with some kindof electric transducer.. pretty awesome. or even just a combination. for example: wind up, release, and settle with a PID or something. better yet, just feed it some equations for a critically-damped harmonic oscillator and you’re done. jump up, down, variable heights and lengths… moving platforms? hehe…
if someone could do this with a leg-a-pod… look out. ;)
@kristian If the system was critically damped I don’t think the controller would have enough gain to perform the jump – it relies heavily on the initial peak time to make it up to the next level. I was saying there is just too much oscillation and the Ts(2%) settling time is too high.
I think the existing design is probably fine, but they need to model the transfer function in the Laplace domain (if they haven’t already) and add an active PD or a more compromising lag-lead compensator to improve the transient response.
@ NatureTM, Re Bose Suspension.
Yep, it was a research project several years ago, I compiled some videos and info for a presentation at least 5 years ago for an Auto Technology class I had.
I believe that it uses a damper for road oscillations over or around 100khz, a torsion bar spring, and a linear electromagnetic coil for the actuator.
They say the system can use a few KW of power, but that during normal driving the power is harvested in damping and used for extending.
Personally I would probably want a proven (and for sale) technology, like the performance airbags I also had in the presentation, they allowed adjustment of both the ride height and the spring rate, so you could drop the car and stiffen the rate for performance driving. I forget the name now, but you can look it up.
It looks strange in the dropdown vid. When the front wheels roll off the platform, they are not dropping down until the rears clear it too.
@ CMJ
Looks odd, because they’re not on independent suspension, but a solid framework holding all 4 wheels. Thats why the front wheels can’t drop until the back wheels have cleared as well.
Now, I wonder if that can be implemented in a bi-pedal design. It’ll be great for a small scale (like 4 – 6 feet tall) mech or something so you wouldn’t have to think up some crazy rocket propulsion system straight out of some sci-fi movie to get it off the ground.
@cde I would even say it’s preprogrammed, cameras would be too complex.