Adaptive Reinforcement Learning Strategy-Based Sliding Mode Control of Uncertain Euler–Lagrange Systems With Prescribed Performance Guarantees: Autonomous Underwater Vehicles-Based Verification

This article studies the tracking control problem of uncertain Euler–Lagrange systems. Despite receiving widespread attention in recent years, the problem remains unresolved to a large content when considering response quality, optimality, robustness, and conservatism. The main challenge lies in how to integrate performance constraints into adaptive dynamic programming (ADP) algorithm and achieve a balance between robustness and conservatism within this framework. To that end, this study proposes a new performance constraint-handling sliding mode manifold, a new prescribed performance function, and a new ADP-oriented observer disturbance observer. The above theoretical findings, together with the fuzzy logic system-based ADP algorithm, realize the convergence of tracking errors to a prespecified residual set within a finite-time setting in an optimal manner, enhance robustness, and reduce conservatism. The proposed controller facilitates the practical application of tracking control for Euler–Lagrange systems. Simulations on an autonomous underwater vehicle demonstrate the effectiveness and benefits of the proposed method.