Abstract: Six degree of freedom collaborative robots are widely used in industrial assembly, precision machining, and medical fields due to their high flexibility and human-machine collaboration characteristics. Error compensation has become the key to improving their motion accuracy. By using the method of comparative argumentation, this paper summarizes the concepts, processes, and steps of three commonly used methods for robot motion error compensation: geometric error compensation, dynamic error compensation, and machine learning based error compensation. Clarified the advantages, disadvantages, and applicable scenarios of three types of robot motion error compensation methods. The geometric error compensation method has low technical and data requirements, which affects its accuracy and applicability; The dynamic error compensation method relies on high-precision sensors and advanced control algorithms to improve applicability, but accuracy is limited by sensor precision; The error compensation method based on machine learning has the highest accuracy and applicability, but requires a large amount of error data for model training. The research results of this article provide a method reference for motion error compensation of industrial collaborative robots.