Abstract

Since UAVs are agile robots that move in higher dimensional space, they have been adopted to solve a diverse range of challenging problems. This talk will introduce the control and challenge of multirotor systems such as distributed formation control, SLAM and motion planning for autonomous UAV, deep learning-based autonomous flight, distributed cooperative control, tracking control of a moving target with unknown motion. The tracking control presents a method based on applying a DNN for detecting a moving target (e.g., car) undergoing unknown motion and without the use of artificial image patterns or tags. The bounding box provides the information required for estimating the target motion based on a kinematics model, and the orientation can be leveraged to control the UAV facing the target at a specified angle. To deal with occlusion problems during tracking, an unscented Kalman filter (UKF) is developed to estimate the unknown position and velocity of the target. Additionally, a high-order polynomial obtained by quadratic programming (QP) is used to emulate the target motion during occlusion so as to facilitate the state estimation in the UKF.

Bio

Prof. Teng-Hu Cheng received his Ph.D. in 2015 from the department of Mechanical Engineering at the University of Florida. His work has been recognized by the 2015 American Automatic Control Council (AACC) O. Hugo Schuck (Best Paper) Award. In 2016, he joined National Chiao Tung University in the Department of Mechanical Engineering. His research interests include robotics, autonomous UAV systems, networked system control, switched control, event-driven control, and nonlinear control.

All faculties and students are welcome to join.