Model predictive control and vision-based control strategies for Quadrotor systems

点击翻译按钮获取中文摘要

Quanser Consulting Inc. has many-year experience of designing and implementing reliable and efficient Unmanned Vehicle Systems. Quanser 's product - QBall 2 quadrotor - is an innovative indoor rotary wing platform suitable for a wide variety of unmanned vehicle research applications, including vehicle modeling and control, motion planning, obstacle avoidance, sensor fusion, fleet maintenance, fault-tolerant control, multi-agent navigation, and so on. A prerequisite for above frequently reported research on the quadrotors is to design and implement a reliable and efficient control system. Some control strategies are already applied on the quadrotor flight; e.g, PID control strategy, backstepping-based nonlinear control and neural networks based control strategies. LQR control strategy on Quanser QBall 2 aims at operating a system with minimum energy consumption. According to the two identified problems from the company, this project will target at the following two tasks during the six-month time-frame: (1) To develop Model Predictive Controller for the Quanser QBall 2, and (2) to develop vision-based controllers for the Quanser QBall 2. By developing novel analysis and synthesis tools, we expect to achieve greatly improved control performance for the quadrotor control systems, which would broaden its range and enhance its performance in industry and teaching applications. New and strong collaborative relationship will be established through this project. The proposed project would fill in the gap between theory and practice; it would provide control engineers with new tools for analysis and synthesis of quadrotor flight control systems; finally, it would benefit the corresponding industry in Canada. Under the supervision of the applicant and the Quanser Design Engineer, two graduate students will receive research training and gain hands-on practical control system design experience from this project.