Feedforward-Feedback Integration in Flight Control: Reinforcement Learning with Sliding Mode Control

TL;DR

This paper introduces a hybrid control framework combining reinforcement learning and sliding mode control to improve aircraft flight control, ensuring robustness and better transient response under input constraints.

Contribution

It develops a novel learning-augmented control method that guarantees stability and robustness while enhancing transient performance in nonlinear, underactuated systems.

Findings

Hybrid controller outperforms standalone RL and SMC in simulations

Improves transient behavior and reduces control oscillations

Maintains robustness under uncertainties and disturbances

Abstract

Learning-based controllers leverage nonlinear couplings and enhance transients but seldom offer guarantees under tight input constraints. Robust feedback like sliding-mode control (SMC) provides these guarantees but is conservative in isolation. This paper creates a learning-augmented framework where a deep reinforcement learning policy produces feedforward commands and an SMC law imposes actuator limits, bounds learned authority, and guarantees robustness. The policy is modeled as a matched, bounded input, and Lyapunov-based conditions link SMC gains to the admissible feedforward bound, guaranteeing stability under saturation. This formulation is applicable to nonlinear, underactuated plants with hard constraints. To illustrate the methodology, the method is applied to a six-degree-of-freedom aircraft model and compared with Reinforcement Learning and isolated SMC. Simulation results show that the hybrid controller improves transient behavior and reduces control oscillations compared to standalone RL and SMC controllers, while preserving robustness under modeling uncertainties and disturbances. Even using it with partially trained policies, SMC component of the control stabilizes transients, whereas fully trained policies provide faster convergence, reduced constraint violations, and robustness. These results illustrate that learning-augmented control offers superior performance with robustness guarantees under tight input constraints.