Practical three-dimensional path following and control for unmanned combat aerial vehicles based on expansion of L 1 + guidance logic and disturbance rejection

Abstract

An effective virtual-target-based three-dimensional path-following algorithm is proposed for unmanned combat aerial vehicles, which extends from the planar $L_1^{+}$ guidance. The separated implementation of the planar guidance law in two perpendicular planes provides the longitudinally-laterally decoupled acceleration commands. In order to obtain a non-unique virtual-target-point, a numerical iterative method using auxiliary path-related variables is proposed. Moreover, two protection improvements are discussed for real-word implementation: one involves the modification of adaptive $L_1^{+}$ length for vehicle position far away from the desired path; the other involves a switching strategy for multi-segment paths. The improved algorithm eliminates complex coordinate transformation and easily connects different types of path segments. For the control loop, technology enhanced with the active disturbance rejection control was used to facilitate the design of a bank-to-turn autopilot. For the pitch and roll channels, a rate-damping augmentation loop was used first to transform the original system into a generalized plant. Then the second-order active disturbance rejection control was structured to estimate the total disturbance, including time-varying dynamics, model mismatch, and coupling of channels, making the system an ideal integral-series type. The non-smooth optimization technique in the H-infinity methodology was used to facilitate tuning of the fixed-structure active disturbance rejection control.