Application of Sliding Mode Theory to Guidance and Control of Unmanned Aerial Vehicles


The main objective of the lateral guidance algorithm is to keep the vehicle on preplanned desired path by controlling the lateral track errors during flight and to keep them as small as possible by generating suitable reference commands. Cross track (lateral) error control of unmanned aerial vehicles (UAVs) in the presence of uncertainties and disturbances with bounded control input (φref ) is a challenging task. The path following guidance law needs to be devised using generalized kinematic model and by explicitly considering the UAV autopilot dynamics. However, the inclusion of these dynamics into guidance design further complicates the problem by increasing the relative degree, and stability, and control boundedness becomes difficult to analyze. To address these challenges, several studies for inclusion of autopilot dynamics into guidance design are presented in this thesis for lateral path following applications.

Firstly, the guidance and control framework based on sliding mode theory is presented to solve the two dimensional path-following problem. Limitations of the existing nonlinear sliding surface for lateral guidance are indicated and thus two novel stable nonlinear sliding manifolds are proposed for the guidance problem. The two surfaces are then employed to generate two new nonlinear guidance laws for UAV path following. The proposed guidance schemes rely on First Order Sliding Mode Control (FOSMC) algorithm derived at the kinematic level generating reference bank commands. The autopilot based on super twisting algorithm using linear sliding surface forms the inner control loop for control actuation.

The autopilot is involved in the feedback nonlinear sliding mode based guidance law design for path following of UAVs. The major contribution of this work is the dynamics of the autopilot taken into account for guidance law design, along-with saturation constraints on guidance commands for high performance in all scenarios. To solve relative degree two problem, a nonlinear sliding manifold is used with real twisting algorithm for guidance design, the guidance loop generates bank angle commands for executing roll maneuvers. The strategy provides a framework to implement the developed controller on the experimental vehicle without modifying the key structure of the original autopilot controller.

Moreover, an innovative sliding mode based partially integrated lateral guidance and control scheme for UAVs is proposed. Guidance and control framework based on second order sliding modes is presented to solve the problem of two dimensional path-following. The main contribution of the technique presented here is the partial integration of the two loops i.e., a guidance and control system via series interconnection of two stable sliding manifolds. The proposed guidance scheme relies on a nonlinear switching surface with the real twisting algorithm derived at the kinematic level, generating roll error commands. The autopilot based on the super-twisting algorithm using a linear sliding surface forms the autopilot loop.

Finally, a new guidance law for accurate following of flight path to observe tight ground track control is presented. The unique feature is to explicitly account for autopilot constraints by defining a 3-D sliding manifold. The guidance solution described is based on state stabilization of kinematics-dynamics trajectories i.e., the guidance law is evolved based on the knowledge of dynamical characteristics of the UAV. A robust FOSMC guidance algorithm is derived using the nonlinear 3-D sliding manifold to develop the guidance law.

For the proposed schemes, proof of existence of sliding mode, actuation boundedness and performance of the path-following closed-loop system is analyzed. Flight results validate the performance and effectiveness of the proposed framework for guidance and control design.

Download full paper