Neural network-based co-simulation for enhanced analysis of 3RRR parallel robot

This study explores the potential of a neural network-based co-simulation procedure in analyzing the dynamics of a parallel mechanism, specifically focusing on the 3RRR robot. First, the application of neural networks to co-simulation is explained. Subsequently, comprehensive formulations for both forward and inverse dynamics are presented, specifically tailored to the benchmark problem at hand. To streamline computational efficiency, the Augmented method to transform the resulting Differential Algebraic Equations (DAEs) into Ordinary Differential Equations (ODEs) is introduced. By evaluating the achieved results against the desired trajectory and the co-simulation response for a 3RRR parallel robot, the remarkable efficacy of proposed method is established. Notably, the integration of co-simulation with neural networks, where driver and driven links are solved individually, enables precise analysis outcomes while significantly reducing computational time. This research unveils an innovative avenue for enhancing the analysis of parallel mechanisms, promising far-reaching implications in the field of robotics and beyond.