Adaptive Neural Network Tracking Control of Flexible-Joint Robots with Guaranteed ℒ∞ Tracking Performance

The focus of this research is to create an adaptive tracking controller for flexible-joint robots (FJRs) that don’t compromise on the robustness of the tracking performance while overcoming the limitations imposed in previous studies. The proposed controller formulates the backstepping NNs (Neural Networks) along with the error term modification in the adaptive design to ensure that tracking errors are controlled within the [Formula: see text] norm. To boost the computational time and to make the control structure easier, DSC (Dynamic Surface Control) and MLP (Minimal Learning Parameter) strategies are employed simultaneously. The simulation and experimental testing on a single-link robotic system indicate that the controller provides accurate tracking performance and keeps the tracking errors bounded even in the presence of system uncertainties and nonlinearities. This research offers a newly developed adaptive control strategy that relieves several restrictions typical of earlier stability analyses of FJRs, hence providing a more efficient and cost-effective solution through the combined use of DSC and minimal learning parameters. Besides, the controller’s effectiveness has been confirmed through its application on a single-link robotic system.