PhD - Ricardo Cajo - Fractional Order Control Strategies for Autonomous Agents

PhD - Ricardo Cajo - Fractional Order Control Strategies for Autonomous Agents

Unmanned Aerial Vehicles (UAVs) and Unmanned Ground Vehicles (UGVs) are two main types of autonomous agents that have awakened academic and commercial interest within the robotics community. They are considered ideal platforms that can work individually and collectively to perform important tasks such as reconnaissance, surveillance, combat, rescue, agriculture, etc.

Many of these tasks in which autonomous mobile robots are involved will have to be performed in complex, unknown, and challenging environments using its limited physical and computational resources. Hence, the control system must ensure in real-time that the robot can achieve its tasks despite external events due to payload variations, manufacturing variations, modeling uncertainties, among others. Besides, the energy consumption optimization to increase the flight time is one of the challenges in UAV systems. Therefore, it is important to develop effective control strategies focusing on trajectory control, energy consumption and disturbance rejection.
The emergence of the fractional calculus has allowed to obtain a better response of closed-loop control systems and a better representation of hidden properties for complex modeling phenomena. In view of the opportunity, this thesis develops different fractional control strategies for single and multiple autonomous agents in a diversity of test scenarios focused on trajectory control, energy consumption, and disturbance rejection.

Firstly, fractional-order proportional derivative (FOPD) controllers are studied for path-following control of the AR.Drone quadrotor in static environments. Two different approaches are proposed to tune the controller parameters.

Secondly, a Fractional Order Model Predictive Control is proposed, where the effect of the weighting factors in the MPC cost function is studied. Lastly, two main control problems that involve multiple autonomous agents based on fractional-control protocols are addressed. Consensus and formation control for multi-agent systems modeled with single and double-integrator dynamics by considering constant disturbances are studied