Quantcast

getting good grasps with mimic joints

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

getting good grasps with mimic joints

Farhan
Hi Rosen,

If I use mimic joints with my robotic hand attempting to grasp a bottle, the OpenRave grasp planner either returns no grasps or poor grasps where the fingers curl in without even touching the bottle. We are using a robotic hand with four fingers and a thumb, where each finger has 1 actuated joint (mcp) and two mimic joints (pip,dip), where the pip joint's equation is 1.1*mcp and the dip joint's equation is .8*dip. The thumb has four actuated degrees of freedom.

If I make all of the joints actuated, I get good grasps in the planner, but the values are infeasible with my actual robotic hand (the mcp joint value is usually around 1.4 and pip is .7, but mechanically, pip/dip have to be greater than mcp).

With my robotic hand, the optimal joint values for a grasp should probably be around 1.0 for the mcp and the rest follows. If the mcp joint angle value is much higher, the fingers curl in too quickly.
 
What do you suggest I do? I've tried setting preshapes and setting artificial joint limits (mcp can't be higher than 1.0) but to no avail.

Thanks!
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: getting good grasps with mimic joints

Farhan
Here is an image that shows the problem.

Loading...