Industrial robot trajectory error compensation based on enhanced transfer convolutional neural networks

Abstract

Abstract This study introduces an enhanced algorithm that integrates transfer convolutional neural networks (CNNs) with radial basis functions (RBFs) to solve the trajectory error problem commonly found in industrial robots. The proposed model utilizes the advanced data mining capabilities of CNNs by combining additional feature extractors and two independent classifiers to support erroneous data analysis. The inclusion of RBFs enhances global accuracy and precision, while mitigating the risk of local convergence. The performance of the proposed algorithm was benchmarked against three other prevalent methods. In 50 iterations, the average accuracy was 85.85% and the F 1 value was 74.3614, demonstrating excellent results. These findings underscore the high applicability and reliability of the proposed method in addressing industrial robot trajectory errors, significantly improving robotic navigation autonomy and intelligence. The results of this study pave the way for future advancements in robot trajectory accuracy and error compensation.