Model-free Adaptive Iterative Learning Control for Nonlinear Systems With Time-varying Constraints

In this paper, an approach is designed to tackle the complex challenges of modeling unknown systems. This approach addresses the difficulty of relying on accurate modeling information when ensuring the security of a dynamics system. A model-free adaptive iterative learning safety control approach is designed based on the iterative control barrier function. Firstly, the dynamic model of the unknown system is discretized to create a discrete-time model. This transformation incorporates nonlinear model uncertainty by employing iterative dynamic linearization. Simultaneously, the uncertainty is estimated and compensated by radial basis function neural networks (RBFNNs). To ensure safety under time-varying constraints, a model-free adaptive iterative learning method with discrete control barrier function is designed based on the designed dynamic iterative model. Ultimately, the validity of the proposed approach is verified on the Franka-Panda robot platform.