It's great given to where we are right now on a technological level, but until we get decent artificial muscles, all biomemetic robots will look like a spastic person having uncontrolled convulsions.
I think you underestimate good engineering. The Festo Smartbird is a good counter example (https://www.festo.com/group/en/cms/10238.htm). I'll agree that we still have a way to go with bipedal robots, but other biomemetric robots (Boston Dynamics) are very impressive these days.
Yes, it can fly very well- but can it fly in place, take off vertically, can it fold its wings - or start from water.
Good engineering is polishing one finite set of movements to a mirror shine here.
Artificial muscles would allow for controll, that allows for all this movements to be realized, without hitting the limitations of the admittetly very well working bionics shown .
I find Festo's work interesting, but the spider bot in particular looks really clunky overall in comparison to even some of the hobby level stuff that's out there. It would make sense to have a rolling robot if it was really fast (and could steer while rolling).
Here's an example of a hexapod from 6 years ago, I believe built from a kit but running customized motion control code:
There are some more recent kits[1] which come close. I wouldn't think it would take a big change to hexapods like these to give the legs a rounder profile to roll on, if they curled their legs like the Festo.
I believe (please correct me if I'm wrong) that the kinematics for those two robots are pre-programmed, and will only work on flat surfaces with no obstacles. Yes, the main body does have an accelerometer which can sense its gross orientation, but if they're just using regular servos, there is no feedback from the legs to indicate height changes in the terrain or obstacles.
Pre-programmed kinematics makes for a cool-looking demo, but isn't going to be super-useful for making an all-terrain robot successful.
Oh absolutely - but GR won't sell me a kit for $300 either... I wonder though if rolling doesn't make proper feedback less important - 'if the surface isn't level, start rolling!'
I'm hoping that GR will have development kits out in the not-too-distant future.
The hardware is "cheap", in the sense that each leg has two motors and that's it. All feedback is via position and torque sensing on the motors themselves. The mechanicals for the legs are also relatively simple and cheap.
It is the software which really makes the system perform.
I love these guys! The Festo robotics group was one of the sponsors for our Battlebot back in the early naughts and unlike the other sponsors who were only concerned with how their logo would look on TV the Festo folks really got into the technical aspects of our designs. They went so far as to modify one of their pan head cylinder designs so that it would fit under the 'wedge' of our robot and allow us to throw another robot out of the ring. Sadly Battlebots had died before we really had a chance to try something like that out.
I particularly like the gait work they have done with the rolling spider robot. Legged robots are good for really rough terrain and crap for efficiently going long distances. It is a good compromise.
17 comments
[ 46.0 ms ] story [ 969 ms ] threadGood engineering is polishing one finite set of movements to a mirror shine here.
Artificial muscles would allow for controll, that allows for all this movements to be realized, without hitting the limitations of the admittetly very well working bionics shown .
https://www.festo.com/cms/en-us_us/15737_15744.htm
Here's an example of a hexapod from 6 years ago, I believe built from a kit but running customized motion control code:
https://www.youtube.com/watch?v=rAeQn5QnyXo
Skip to 0:45 to see it running. Still amazing to me and I haven't seen too many other legged robots that look this smooth.
1: https://youtu.be/-Dt2yVR3FJQ
Pre-programmed kinematics makes for a cool-looking demo, but isn't going to be super-useful for making an all-terrain robot successful.
Compare that to what Ghost Robotics is doing:
https://www.youtube.com/watch?v=bnKOeMoibLg
In addition to the other capabilities shown, at the very end of the video it shows how quick and accurate the force feedback is.
And the quick reflexes on the ice are also pretty impressive.
The hardware is "cheap", in the sense that each leg has two motors and that's it. All feedback is via position and torque sensing on the motors themselves. The mechanicals for the legs are also relatively simple and cheap.
It is the software which really makes the system perform.
I particularly like the gait work they have done with the rolling spider robot. Legged robots are good for really rough terrain and crap for efficiently going long distances. It is a good compromise.
quite similar to this flying fox.