Hi there,
I’m currently studying for an MSc in Robotics and for my dissertation project I’m building a robot to perform some manipulation tasks using a Lynxmotion robot arm.
As part of this project I’m trying to use a camera to estimate the joint angles of the arm to hopefully obtain more accurate control.
Leaving aside the technical difficulties (of which I know there are many) of getting the vision side to work. I’d like to get people’s opinions on how useful they think that this would be and check that my understanding of servos is correct (as I’m mainly coming at this from a software point of view), before I head off down a blind alley.
Anyway,
-
Would it be possible to get more accurate arm movements if the actual joint angles were known? From video clips I’ve seen it seems as if the arms are already fairly accurate for pick and place type activities so the main use I see for this would be in determining if the arm has hit an obstacle, automatic gravity compensation when picking up objects of different weight and possibly for auto calibration.
-
Am I correct in my understanding, that although the arm servos contain an internal potentiometor for closed loop positioning, there’s no way for control software to actually read the joint values (short of physically modifying the servo)?
-
I notice that the SSC-32 offers speed control of the servos and also via RIOS, acceleration/deceleration control. How is this done? I thought with the servo motors you could only set the desired angle.
-
In a similar vein, how does the RIOS gravity compensate feature work? When providing a higher torque, does it just try to drive the servo to angle past where it thinks it needs to go?
Thanks in advance for you time, and please be tolerant if I’ve asked any really dumb questions. 
Regards
Alan