Learning Control of an Overhead Crane for Obstacle Avoidance - American Control Conference, 1997. Proceedings of the 1997

A control strategy is presented for a 3-dimensional overhead crane. Using the differential flatness of the crane and a parameterization method, we calculate the optimal trajectory of the payload that results in minimum transfer time when there exist obstacles in the direct moving path. A learning algorithm is then used to generate the desired feedforward input signal that can drive the system output to track the optimal trajectory exactly.