Les commentaires sont clos.

Developing advanced predictive tracking controllers for quadrotors

28 Avril 2023

Catégorie : Doctorant

The aim of the PhD thesis is to develop accurate and robust controller for autonomous navigation of mobile robots and target tracking using visual information.


Host Laboratory: ImViA - VIBOT Team in Le Creusot, Université de Bourgogne

Keywords : Unmanned Aerial Vehicles, Model Predictive Control, Visual Predictive Control, Visual Servoing, Mobile Robotics.


In recent years, quadrotor unmanned aerial vehicles have attracted increasing attention from both industrial and academic communities. With characteristics such as vertical take-off and landing, single or team flight, and low-cost manufacturing, quadrotors have broad application value in military strategy [1], disaster rescue [2], tracking and shooting [3], and monitoring and recognition [4]. Quadrotor UAVs are underactuated and strongly coupled nonlinear systems, subject to structural uncertainties and unknown external disturbances. Therefore, designing an accurate and robust controller for the quadrotor to achieve autonomous flight and target tracking is a great challenge.

Recently, with the increasing development of fast computers, model predictive control approaches (MPC) have become real-time applicable for nonlinear mechatronic systems. MPC refers to a set of controllers that use a model to compute inputs from the current time to a future time to optimize the behavior of a model along the input trajectory. The predictive nature of the control design makes it ideal for high-performance trajectory tracking. A key advantage of MPC is that it offers the ability to design controllers with constraints while solving an optimal control problem along the given trajectory. Because of its efficiency and advantages, the predictive control strategy is regularly used to control robotic systems such as quadrotors [5]. In other words, predictive control has also been extended to the case of visual servoing (VS), giving rise to a new approach named visual predictive control (VPC) [6].

The development of Model Predictive Control (MPC) has provided the essential background to formulate Visual Servoing (VS) as a constrained optimization problem. The main objective of Visual Predictive Controllers (VPCs) is to provide a systematic framework to accomplish VS problems in a mathematically optimal fashion while considering the inputs, states, and task constraints into account. In most of the proposed VPC approaches, constraints due to the camera's Field of View (FoV), the kinematics of the robot, and sensor/input saturation are considered. Despite this, no systematic strategy has been provided in the above-mentioned literature to deal with system/measurement uncertainties. Although VPC gives satisfactory performance in tracking a target, it suffers from some inconveniences due to various factors, including imperfect system models, measurement noises, and exogenous disturbances such as wind. Also, some uncertainties can arise from the VS system, such as having a parametric/deterministic nature (e.g., kinematics error in the quadrotor model or calibration error in the focal length of the camera).

The thesis work will be based on a particular application framework, namely, a fast-moving target for a quadrotor subject to many uncertainties, such as uncalibrated camera, wind disturbance, uncertain parameters, and measurement noise. Very few works can currently be identified for this type of application [7].

The work plan for this thesis will take place in the following chronological order:

  • Conduct a bibliographic study of visual predictive control and of robust control applied to robotic systems.
  • Develop an intelligent method to find a visual feature representation that is robust to big dynamic transformation movements and is suitable to be an optimization variable.
  • Study the visual predictive control strategy for quadrotor that allows the tracking of reference trajectory.
  • Develop and implement an approach that allows the quadrotor to follow a moving target despite the existence of uncalibrated camera, wind disturbance, and measurement noise.
  • Implement the developed advanced predictive controller in a drone (AR-drone, Quanser drone).
  • Compare the results and evaluate the performance.

Bibliography :

[1] Wang, L.;Wang, J. Four rotor UAV Special Ammunition Hovering launch Mechanics Research. J. Ballist. 2022, 34, 38–44.

[2] Wang, W. Multirotor UAV design and application in fire fighting and rescue operations. Electron. Compon. Inf. Technol. 2022, 6, 185–187.

[3] He, F. Pedestrian detection and route tracking from aerial view of quadrotor UAVs. Electron. Meas. Technol. 2022, 45, 50–56.

[4] Wang, S. Multirotor drones in the field of public safety application idea. Robot. Ind. 2022, 3, 36–41.

[5] Aliyari, M., Wong, W. K., Bouteraa, Y., Najafinia, S., Fekih, A., & Mobayen, S. (2022). Design and Implementation of a Constrained Model Predictive Control Approach for Unmanned Aerial Vehicles. IEEE Access, 10, 91750-91762.

[6] Jin, Z., Wu, J., Liu, A., Zhang, W. A., & Yu, L. (2021). Gaussian process-based nonlinear predictive control for visual servoing of constrained mobile robots with unknown dynamics. Robotics and Autonomous Systems, 136, 103712.

[7] Zhang, K., Shi, Y., & Sheng, H. (2021). Robust nonlinear model predictive control based visual servoing of quadrotor UAVs. IEEE/ASME Transactions on Mechatronics, 26(2), 700-708.

Applicant profile:

  • Master or Engineering degree in computer vision and robotics or related fields.
  • Good knowledge of control systems as well as image and signal processing.
  • Solid background in mathematics, simulation and coding in MATLAB, SIMULINK, Python, C/C++, and ROS.
  • High motivation for theoretical and practical research work.
  • Good research methodology and very good skill in scientific English writing.
  • Great interest in disseminating research results through publications and presentations at international conferences.

Finacial support and deadlines

Financing Institution: MESRI Scholarship for 36 months.

Application deadline: 28 May 2023 at 23h59.

Start of contract: October 2023.

Thesis Supervisor(s) :

  • Amine ABADI,


Applicants are invited to submit their application to the PhD Thesis supervisors. Application must contain the following documents:

  • Cover letter describing your training, experiences, knowledge and most significant contributions, and how you can contribute to this project.
  • CV.
  • Copy of academic transcripts (Bachelor and Master/Engineering).
  • Copy of certificates of successful completion or diplomas of the Bachelor and Master/Engineering degrees.
  • At least one reference letter.
  • Any other relevant information (webpage, github, publications, …).