Application of Newton–Euler Algorithm Based Dynamics Control Technology for SCARA Robot

Abstract

Aiming at the bottlenecks of traditional SCARA robot dynamics control method, such as high computational complexity, insufficient parameter identification accuracy, and weak anti-interference ability, a recursive Newton–Euler control framework based on genetic algorithm optimization is proposed. The optimal performance of the Newton–Euler method could achieve 98% accuracy, which was 7%–10% higher than that of the PSO/machine learning model. The NEA recursive computing architecture was designed to reduce the dynamic analysis complexity of the multi-joint system from O(n³) to O(n), and the single-cycle computation time was reduced to 9.0 s (efficiency increased by 14.3%). The practical test results showed that the discrimination rate of the dynamic parameters of the robot based on the model was higher than that of the dynamic control model based on machine learning, which could reach more than 90%. The stronger the stability, the smaller the torque change caused by the collision between the robot and the object, and the variation range was from 90 to −30 nm. In conclusion, the SCARA dynamic control model based on the Newton–Euler algorithm has high control accuracy and stability. The research breaks the contradiction between precision and real time in highly dynamic scenes and provides a new paradigm for the precision control of industrial robots. In the future, reinforcement learning will be integrated to build a hybrid architecture to improve the adaptability to complex working conditions.