In recent years, the use of deep neural networks as function approximators has enabled researchers to extend reinforcement learning techniques to solve increasingly complex control tasks. The emerging field of deep reinforcement learning [1] has led to remarkable empirical results in rich and varied domains like robotics [2], arcade games [3], and multiagent interaction [4]. This seminar will help students gain a high-level view about the current state of the art and potential directions for future research [5]. Participants are expected to actively participate in the course by presenting selected articles [6] and discussing code implementation of the deep learning architectures, e.g. [7-18] etc.
[1] https://deepmind.com/blog/deep-reinforcement-learning
[2] https://blog.openai.com/learning-dexterity
[3] https://youtu.be/XjsY8-P4WHM?t=48s
[4] https://www.pommerman.com
[5] https://www.dropbox.com/s/99eyutemrdb17kj/SIAM%202018.pdf
[6] https://rebrand.ly/DRL
[7] https://github.com/xbpeng/DeepMimic
[8] https://github.com/MarkPKCollier/NeuralTuringMachine
[9] https://github.com/brendenlake/BPL
[10] https://github.com/deepmind/dm_control
[11] https://github.com/atenpas/gpd
[12] https://github.com/ros-planning/moveit
[13] https://github.com/PointCloudLibrary/pcl
[14] https://github.com/cbfinn/gps
[15] https://github.com/Unity-Technologies/obstacle-tower-env
[16] https://github.com/tensorflow/models/tree/master/research
[17] https://github.com/MarcToussaint/18-RSS-PhysicalManipulation
[18] https://www.youtube.com/watch?v=-L4tCIGXKBE
[etc.]
CNN-SLAM: Real-time dense monocular SLAM with learned depth prediction
https://www.youtube.com/watch?v=z_NJxbkQnBU
DSO: Direct Sparse Odometry
https://www.youtube.com/watch?v=C6-xwSOOdqQ
ORB-SLAM2: an Open-Source SLAM for Monocular, Stereo and RGB-D Cameras
https://www.youtube.com/watch?v=ufvPS5wJAx0
Autonomous Drone Navigation with Deep Learning. Flight over 250 meter Forest Trail
https://www.youtube.com/watch?v=H7Ym3DMSGms
AirSim Demo
https://youtu.be/-WfTr1-OBGQ
https://github.com/Microsoft/AirSim
DeepLoco: Highlights
https://youtu.be/G4lT9CLyCNw
Michiel van de Panne
https://www.cs.ubc.ca/~van/papers/index.html
Ingredients for Robotics Research
https://youtu.be/8Np3eC_PTFo
https://openai.com/blog/ingredients-for-robotics-research
https://gym.openai.com/envs/#robotics
OpenAi Baselines >> HER
https://github.com/openai/baselines
https://github.com/openai/baselines/tree/master/baselines/her
Contact-Invariant Optimization for Hand Manipulation
https://www.youtube.com/watch?v=Gzt2UoxYfAQ
Emo Todorov
https://homes.cs.washington.edu/~todorov/projects.html
Xperience.org
http://www.xperience.org/index.php/publications.html
Deep Q-Network & Dueling network architectures for deep reinforcement learning
https://youtu.be/XjsY8-P4WHM
SIPB Deep Learning Group
https://github.com/pmiller10/cambridge-ai
YOLO COCO Object Detection
https://youtu.be/yQwfDxBMtXg
YOLO Algorithm
https://youtu.be/9s_FpMpdYW8
Rhythmus | Tag | Uhrzeit | Format / Ort | Zeitraum |
---|
Modul | Veranstaltung | Leistungen | |
---|---|---|---|
39-Inf-EGMI Ergänzungsmodul Informatik | vertiefendes Seminar 1 | unbenotete Prüfungsleistung
|
Studieninformation |
vertiefendes Seminar 2 | unbenotete Prüfungsleistung
|
Studieninformation | |
vertiefendes Seminar 3 | unbenotete Prüfungsleistung
|
Studieninformation | |
vertiefendes Seminar 4 | unbenotete Prüfungsleistung
|
Studieninformation |
Die verbindlichen Modulbeschreibungen enthalten weitere Informationen, auch zu den "Leistungen" und ihren Anforderungen. Sind mehrere "Leistungsformen" möglich, entscheiden die jeweiligen Lehrenden darüber.