| 摘要 |
PROBLEM TO BE SOLVED: To provide a robot control method by which a track error can be reduced, and further a learning speed can also be reduced.SOLUTION: An evaluation value E, in which a track error ebetween a target track x and a real track yis evaluated, is calculated. When the calculated evaluation value Eis better than a best evaluation value E, the calculated evaluation value is renewed and stored as the best evaluation value E, and a command track uon this occasion is renewed and stored as a best command track u. When the calculated evaluation value Eis worse than the best evaluation value E, a compensator, which calculates a correction track &Dgr;u, is exchanged with another compensator, the correction track &Dgr;uis calculated, and a next correction value uis calculated on the basis of the aforesaid correction track &Dgr;uand the best correction track u. |
