Error identification and compensation in large manipulators with application in cancer proton therapy

Abstract

Important robotic tasks could be effectively performed by powerful and accurate manipulators. However, high accuracy is generally difficult to obtain in large manipulators capable of producing high forces due to system elastic and geometric distortions. A method is presented to identify the sources of end-effector positioning errors in large manipulators using experimentally measured data. The method does not require explicit structural modeling of the system. Both geometric and elastic deformation positioning errors are identified. These error sources are used to predict, and compensate for, end-point errors as a function of configuration and measured forces, improving the system absolute accuracy. The method is applied to a large high-accuracy medical robot. Experimental results show that the method is able to effectively correct for the system errors.