High-performance position control can be improved by the compensation of gear backlash. This paper presents a compensation method that does not need an output position sensor. In the first section, two different methods of on-system backlash identification are described. One method is based on dynamic reaction to a small torque impulse injected by the motor, while the other analyses the change of torque at the end of the backlash dead zone in a quasi-stationary trajectory. The second section of the paper deals with backlash compensation. Based on the identified value of backlash, the angular error of the output shaft is reduced. A self-learning time-optimal control method is described. Experimental measurements demonstrate the excellent performance and accuracy of the compensated system.
展开▼