Intendierte Lernergebnisse
The focus of this lecture is to provide a detailed overview of how visual perception is used in robotics to derive information of the environment a robot is operating in.The class dives into the full image processing pipeline - from image formation to advanced computer vision methods, such as i.e. 3D perception, motion estimation, and scene reconstruction, with examples from autonomous robotic systems on Earth and in Space. An additional aspect will be an introduction on how to simulate vision sensors to enable the evaluation of image-based algorithms.At the end of the lecture, students will have a deep understanding of how images from on-board cameras are used in robotic visual perception.An opportunity to apply the acquired knowledge in selected research projects is provided by the accompanying practical class “Hands on Robotics: Visual Perception”.
Lehrmethodik
Additional to the material presented in the lecture, a literature collection including research papers and books will be provided for students who are interested in extended study.It is suggested to take the accompanying practical course (Labor) to this lecture. Both are strongly linked such that the learned theory can immediately put into practice.
Inhalt/e
Image formation – biological vision and camera systemsLow-level image processingVisual features, feature detection and feature matching2 view geometry and outlier rejectionStereo and multi-view vision for 3D perceptionVisual odometry and SLAM systemsMappingObject detection/recognitionModelling vision sensors in simulation and computer graphics
Erwartete Vorkenntnisse
Good knowledge of Mathematics (linear algebra).Basic knowledge in Robotics
Literatur
Books:Richard Szeliski: Computer Vision: Algorithms and Applications, 2nd ed. Paper or online.David A. Forsyth: Computer vision: A modern approach. Paper or online.Link auf weitere Informationenhttps://campus.aau.at/studium/course/118550