Active vibration control of flexible-joint manipulators using accelerometers

Purpose Manipulators are often subjected to joint flexibility caused by various causes in industrial applications, such as shaft windup, harmonic drives and bearing deformation. However, many industrial robots are only equipped with motor-side encoders because link-side encoders and torque transducers are expensive. Because of joint flexibility and resulted slow response rate, control performance of these manipulators is very limited. Based on this, the purpose of this paper is to use easy-to-install and cheap accelerometers to improve control performance of such manipulators. Design/methodology/approach First, a novel tip-acceleration feedback method is proposed to avoid amplifications of approximation errors caused by inversion of the Jacobian matrix. Then, a new control scheme, consisting an artificial neural network, a proportional-derivative (PD) controller and a reference model, is proposed to track motor-side position and suppress link-side vibration. Findings By using the proposed tip-acceleration feedback method, each link's vibration can be suppressed correlatively. Through the networks, smaller motor-side tracking errors can be obtained and unknown dynamics can be compensated. Tracking and convergence performance of the network-based system can be improved by using the additional PD controller. Originality/value The originality is based on using accelerometers to improve link-side vibration suppression and control performance of flexible-joint manipulators. The previously used methods need expensive link-side sensors or accurate robot model, which is unavailable for many industrial robots only equipped with motor-side encoders. The report proposed a novel acceleration feedback method and used networks to solve such problems.