Project-Based Learning Using a Quadrotor Testbed

In this paper, we present and discuss the experience we gathered integrating project-based learning in a graduate-level optimal control course at the University of Houston (UH) to support student development. The project consisted of the design, implementation, and experimental validation of an optimal controller for a reference following task using an underactuated multi-rotor unmanned aerial vehicle (UAV) testbed as the plant. We compared the performances of a Linear Quadratic Tracker (LQT) with those of a traditional architecture consisting of cascaded Proportional-Integral (PI) controllers, both in simulation and on a real-world 2 Degrees-of-Freedom (DoF) platform. The educational objectives of the project were manyfold: first, the experiential activity provided students with the opportunity to develop independent learning skills, second, it facilitated a deeper understanding of core optimal control principles both as stand-alone concepts and in comparison with classical control methods, third, it allowed the students to perform the synthesis of controllers from the modeling step to the experimental validation, and fourth, it promoted the acquisition of hands-on "engineering ability". In this contribution, we study the role of the experiential learning intervention that we designed on students’ experiences, engagements, and content knowledge.