Manipulator

Week 1

Week 2

Week 3

Week 4

Week 5

Week 6

Week 7

Week 8

• Using independent PID joint control noticed 10+ mm deviation in end effector Z position... with the accordian suction cup we can only afford +/- 3 mm
• Created dynamic model for the two longer links and implemented time varying LQR to stabilize the trajectory and infinite horizon LQR to stabilize the end point.
• LTV-LQR helped remove oscillations that were present in the indepent joint controller but the dynamic model wasn't nearly accurate enough to remove the steady state error when the arm is extended.
• Implemented radial basis function estimator that learned the feed forward control for the two links as a function of their joint positions.
• With the feed forward estimator we can get the end effector to position within +/- 5 mm. In order for the arm to position any more accurately we need to increase the amount of available torque in one of the motors.

Week 9