Vision and control for autonomous UAVs
A wide variety of UAV tasks involve detecting and tracking an object, and then controlling a desired UAV trajectory relative to this object. Examples include following a person, video inspection of an object in the environment, etc. This project is for a team of at least two students, one working on detection, and the other on control.
For example, detection can be enhanced by using the highly nonlinear model of the drone together with a homography mapping to translate the 3D dynamics of the drone into the 2D image frame. Moreover, an Extended Kalman filter or other nonlinear filter will be used for tracking and sensor fusion between the vision information and data from the inertial measurement unit.
On the control side, the aim is to design a trajectory-following controller that takes as input the detection result. This can be done either directly with visual servoing on the image, or otherwise by estimating the state of the quadrotor in the real world and controlling it with nonlinear design techniques.
This project is connected to an ongoing cooperation with Siemens Germany. Interested students should get in touch one of the contact persons below to setup a meeting.