Re:Father builds exoskeleton to help wheelchair-bound son walk
by jcun4128 on Tuesday July 27, 2021 @02:59PM
(#27976168)
> bike brake cable
Funny you mention that, this guy [1] made an actuator/arm recently doing that technique. Does work but accuracy/flex can be a problem.
Re:Father builds exoskeleton to help wheelchair-bound son walk
by parsecs on Tuesday July 27, 2021 @05:35PM
(#27977957)
Curious, could one apply fancy control theory and modern computing at the problem and fix most of the accuracy and flex problem? Position feedback with integrated sensors can be very compact and highly accurate, and human muscle seem to also use feedback control for holding position..
Re:Father builds exoskeleton to help wheelchair-bound son walk
by aeneasmackenzie on Wednesday July 28, 2021 @12:19PM
(#27985600)
Rather than control theory you can use geometry to reduce the problems with cable driven robots. There are more but I like the Ambidex: https://youtu.be/aLaqMreVj9o
Re:Father builds exoskeleton to help wheelchair-bound son walk
by tremon on Wednesday July 28, 2021 @05:40AM
(#27981935)
The human system doesn't use single-muscle feedback for holding position though; it requires flexing both the push and pull muscles and coordinating the strength between them; positional information is relayed not just from the muscles, but the surrounding skin as well.
Re:Father builds exoskeleton to help wheelchair-bound son walk
by TheAdamAndChe on Wednesday July 28, 2021 @06:54AM
(#27982371)
The constant muscle tone is largely to help bones stay in ball-and-socket joints. Presumably, such tone wouldn't be necessary with other kinds of joints.
Re:Father builds exoskeleton to help wheelchair-bound son walk
by arpafaucon on Wednesday July 28, 2021 @09:42AM
(#27983620)
Hey ! I am not a robotics expert, but I work at this company (Wandercraft) as software dev, so I know some of the reasons :)
First of all I'd say that's because it simpler: we have so much to do that simplicity is a strong decision factor. That way you also reduce the number of moving parts, so maintenance is easier.
On the software side you also get the marginal benefit that you keep a 1:1 mapping between joints and actuator (an example with coupled actuators: Cassie, from the Oregon State University). That way you can have different control strategies for the different joints.
Why do these designs always have the actuators mounted directly at the joints?
Why not have some high modulus yarn transfer the joint loads (ala a bike brake cable, say with high modulus aramid?)? One could keep the actuators near the center of mass (or a trolley that follows it around), and keep the suit fairly minimal.