Semantische 3D-Perzeption für robotische Systeme

Dauer4 SWS
ArtVorlesung mit Übung
SemesterWS2018/19
VortragendeDr. Ing. Michael Suppa
ÜbungsleitungFerenc Balint-Benczedi,Jan-Hendrik Worch
SpracheDeutsch und English
VoraussetzungEchtzeit-Bildverarbeitung
Termine Vorlesung und Übungen18.10.2017-01.02.2018 10:00-14:00
Ort TAB-Gebäude (Am Fallturm 1), Eingang E, Raum 0.36
Wichtige BemerkungenCourse and exercise alternate. Check Schedule below

Pleas sign up for the course on StudIP

Description

Dr. Michael Suppa is a leading expert in robot perception and visiting professor at the University of Bremen. He worked as project manager and researcher at the Institute of Robotics and Mechatronics of the German Aerospace Center (DLR) in the research areas: robotic exploration, 3-D vision, and data fusion. From 2009 until 2015, he was the Head of the Department Perception and Cognition, a recognized world leader on key robotic research topics such as complex scene analysis, perception for resource-limited systems and robotic cognition. In March 2015 he co-founded Roboception, a DLR spin-off company devoted to advancing the State-of-the -Art in 3D sensors and vision.

The lecture covers all important aspects of 3D semantic perception with a high focus on applicability and real world tasks. Some of the topics covered by the lecture:

  • Introduction to 3D
  • Features and Descriptors
  • From features to semantics
  • Scene registration – RGBD SLAM
  • Handling uncertainty in 3D
  • Perception Systems
  • Convolutional Neural Networks
  • and more

Assignments Schedule:

Assignment nr.Handed outSubmission date
A125th October12th November 08:00 AM
A28th November26th November 08:00 AM
A329th November7th January 08:00 AM
A410th January29th January 12:00 PM

Three out of the four assignments must be submitted on time to enter to oral exam. Assignments will not be graded, but they will form a basis for discussion during the oral exam

Course Schedule (subject to change):

The order and date of exercises is subject to change. Check back regularly.

DateActivity
18.10 10:00-11:30 and 12:30-14:00Lecture: Introduction to 3D and Acquisition of 3D data
25.10 10:00-11:30Exercise: Introduction to 3D
01.11 10:00-11:30 Lecture: Features
08.11 10:00-11:30Exercise: 2D Keypoints and descriptors
15.11 10:00-11:30Exercise: 3D Keypoints and descriptors
22.11 10:00-11:30 and 12:30-14:00 Lecture: Objects I and II
29.11 10:00-11:30Exercise: Object recognition
06.12 10:00-11:30Exercise: Discussion of assignment
20.12 10:00-11:30 and 12:30-14:00 Lecture: Scenes I and II
10.01 10:00-11:30Exercise: Segmentation of kitchen scene
17.01 10:00-11:30 and 12:30-14:00 Lecture: CNNs and Applications
24.01 10:00-11:30Exercise: Exercise using CNNs
31.01 10:00-11:30Exercise: Q&A