Skip to Main content Skip to Navigation
New interface

Robust planning and control of unmanned aerial vehicles

Abstract : The objective of this thesis is to realize the modeling, trajectory planning, and control of an unmanned helicopter robot for monitoring large areas, especially in precision agriculture applications. Several tasks in precision agriculture are addressed. In pest surveillance missions, drones will be equipped with specialized cameras. A trajectory will be researched and created to enable unmanned aircraft to capture images of entire crop areas and avoid obstacles during flight. Infected areas will be then identified by analyzing taken images. In insecticides spraying, the aircraft must be controlled to fly in a pre-programmed trajectory and spray the insecticide over all the infected crop areas.In the first part, we present a new complete coverage path planning algorithm by proposing a new cellular decomposition which is based on a generalization of the Boustrophedon variant, using Morse functions, with an extension of the representation of the critical points. This extension leads to a reduced number of cells after decomposition. Genetic Algorithm (GA) and Travelling Salesman Problem (TSP) algorithm are then applied to obtain the shortest path for complete coverage. Next, from the information on the map regarding the coordinates of the obstacles, non-infected areas, and infected areas, the infected areas are divided into several non-overlapping regions by using a clustering technique. Then an algorithm is proposed for generating the best path for a Unmanned Aerial Vehicle (UAV) to distribute medicine to all the infected areas of an agriculture environment which contains non-convex obstacles, pest-free areas, and pests-ridden areas.In the second part, we study the design of a robust control system that allows the vehicle to track the predefined trajectory for a dynamic model-changing helicopter due to the changes of dynamic coefficients such as the mass and moments of inertia. Therefore, the robust observer and control laws are required to adopt the changes in dynamic parameters as well as the impact of external forces. The proposed approach is to explore the modeling techniques, planning, and control by the Takagi-Sugeno type technique. To have easily implantable algorithms and adaptable to changes in parameters and conditions of use, we favor the synthesis of Linear Parameter Varying (LPV) Unknown Input Observer (UIO), LPV quadratic state feedback, robust state feedback, and static output feedback controllers. The observer and controllers are designed by solving a set of Linear Matrix Inequality (LMI) obtained from the Bounded Real Lemma and LMI regions characterization.Finally, to highlight the performances of the path planning algorithms and generated control laws, we perform a series of simulations in MATLAB Simulink. Simulation results are quite promising. The coverage path planning algorithm suggests that the generated trajectory shortens the flight distance of the aircraft but still avoids obstacles and covers the entire area of interest. Simulations for the LPV UIO and LPV controllers are conducted with the cases that the mass and moments of inertias change abruptly and slowly. The LPV UIO is able to estimate state variables and the unknown disturbances and the estimated values converge to the true values of the state variables and the unknown disturbances asymptotically. The LPV controllers work well for various reference signals (impulse, random, constant, and sine) and several types of disturbances (impulse, random, constant, and sine).
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Wednesday, March 24, 2021 - 4:08:11 PM
Last modification on : Thursday, January 27, 2022 - 3:03:45 AM


Version validated by the jury (STAR)


  • HAL Id : tel-03179970, version 1


The Hung Pham. Robust planning and control of unmanned aerial vehicles. Automatic Control Engineering. Université Paris-Saclay, 2021. English. ⟨NNT : 2021UPASG003⟩. ⟨tel-03179970⟩



Record views


Files downloads