Change jogging drift dimensions

[AndyZe]

Very similar to PR #61

#61 defines a service to set velocities in certain dimensions to 0.

This PR defines a service to allow the robot to drift in a smooth but unregulated way (in the specified dimensions). In preparation to take advantage of task redundancy.

[davetcoleman]

Does this support drifting in non-axis aligned vectors / directions?

[AndyZe]

Does this support drifting in non-axis aligned vectors / directions?

Not as I imagine the implementation currently. I can imagine how you would achieve that but it's beyond the scope of what I was hoping to add.

Edit - you could achieve that by adding a new tf frame and setting that frame as the jog_arm control frame. So yes, that could be done.

[davetcoleman]

If instead of bool, we used float, wouldn't that allow more control over how much in each direction you could float? And maybe accomplish arbitrary angles? I'm not really sure, but requiring axis-aligned seems short-sighted for this moveit_msgs API

[AndyZe]

If instead of bool, we used float, wouldn't that allow more control over how much in each direction you could float?

Well it's a binary thing. Either the Jacobian row is present or it isn't, so I don't think a float would help.

The way you get a different axis would be transforming the Jacobian.

[AndyZe]

Does this support drifting in non-axis aligned vectors / directions?

I think this would be implemented better by adding a function to change the jog_arm command frame at runtime.

[davetcoleman]

Could we add a frame (pose) to this message that would support non-axis aligned? Even if you don't use it yet (we could note its not implemented yet)

[AndyZe]

How about adding a geometry_msgs/Transform, and we can assume it will be relative to the jogging command frame?

[davetcoleman]

sgtm

[davetcoleman]

@bmagyar your thoughts are welcome!