While the pages of science fiction novels are saturated with robotic creations sporting a diverse array of ambulatory designs — from skittering, crab-like holoprojectors to robotic serpents — real world robots have historically been somewhat more traditional in their methods of locomotion.
Recent years have brought us a host of novel robotic designs, incorporating everything from snake-like swimming motions to that ubiquitous hallmark of malevolent robotdom: the prehensile metal tentacle.
Now some researchers at Carnegie Mellon have developed a project that tests the feasibility of using a single large ball as a source of locomotion in an effort to “…evaluate the efficacy of this type of dynamic locomotion in the context of human environments.”
The project site offers some intriguing insight into the project and the team, and is well worth a look. (Source: News.com, Carnegie Mellon project site) – [Jeff James]
This just isn’t that cool.
This isn’t the first inverse mouse-ball drive, for one thing. So there’s nothing really special about that.
They use an OTS IMU, which takes away most of the challenge.
And the control system is simple LQR. They added a tiny experimentally-tuned PI loop to correct for frictional effects. This 2D inverted pendulum controller might as well be a homework assignment for a masters level controls class. In fact, I’m not even sure if it is complex enough for that. Maybe it would be best as an undergraudate student thesis.
Oh, and I didn’t even mention that in its current state it can only track a path on carpet, and the only paths it works well with are straight lines; it can do disturbance rejection more robustly though. This makes sense because none of their models involve the dynamics of the input. They built a stabilizer. A 2D inverted pendulum stabilizer. Way to go CMU.
There’s no contribution here. The ballbot just isn’t that cool.
Maybe not, but it does seem as though it could be somewhat usefull. If they made it a little shorter, they could use it as a fancy self-moving serving tray. But realy, if they want it to work as a personal asistant fo the disabled, they should probably make it climb stairs, since not everywhre has elevators andor ramps. And I don’t think a ball-bot would be able to climb that well, unless they make it able to jump or something.
…Actualy, this might be pretty usefull once they give it arms and stuff, but they’ll need to either give it some legs it can pull out when it’s on rough terrain or stair or something, or put two of them together so they can act as a pair of legs. That would probably be pretty cool, realy. Unless you had to clean the rods all the time like the ones in mice.
@ MDude
I think that the ball would do reasonably well on rough terrain… That is if they get a nice tough ball and leave it somewhat deflated, allowing for better traction. I would imagine that if it is deflated enough and it has enough traction that it could climb up stairs.
Jason
The Mindstorms NXT Review