Dynamics based control of vertically articulated manipulators

The relationship between tracking accuracy and the model used in dynamics-based control of a vertically articulated manipulator has been experimentally evaluated. High gear ratios reduce the payload sensitivity and computational complexity of the system dynamics but do not eliminate the performance enhancement from accurate compensation of dominant dynamics. Robot controllers must compensate for changes in dominant drive system, link, and payload dynamics to achieve high speed trajectory tracking accuracy. Feedforward dynamic compensation reduced the peak trajectory tracking errors of an independent-joint PD (proportional differential) controller by a factor of three. Knowledge of payload mass and centroid allowed feedforward dynamic compensation to overcome the disturbances produced by unmodeled payload dynamics. The combination of high sample rates, constant PD gains, and adaptive feedforward dynamic compensation produces a control algorithm with excellent payload independent, peak, and final error