[BasicCartesianControl] needs improvements in movl

[jgvictores]

[BasicCartesianControl] needs improvements in movl.

• Trajectories in simulation need gain and shouldn't need any. Trajectories on the real robot need more gain and are rather faulty.
• Should be able to set velocity instead of duration.

[PeterBowman]

These improvements could be handled by a new controller interface, see #105 (comment):

I was thinking about a new interface (inheritable by BasicCartesianControl and the like) for managing controller configuration during runtime. In fact, it would be nice to have the ability to select a different gain, tweak the maximum joint velocity and trajectory time (along the lines of #46), change the trajectory generator, enable RPC/streaming mode, or even choose the reference frame for input coordinates. Do you think it's feasible?

[PeterBowman]

We are going to dismiss the proposal quoted in my previous comment and implement new methods directly in the controller interface. Until that, this issue is blocked by #121.

[jgvictores]

Related: #135

[PeterBowman]

It's almost two years now and nothing has moved here. I guess this issue originated when the first implementation for BasicCartesianControl became available and still required testing. Conclusion:

• The first subtask may be deemed invalid (either unclear or solved). I've been working with this controller on a real robot and didn't notice any particular issues with the controller gain. Also, I can set it to zero and still work on the simulator.

• The second subtask has been moved to Make linear trajectories depend on velocity instead of time #157.