Munich Datageeks - October Edition

It is time for our next Meetup. The Volkswagen Data Lab will host us once again.

Format:

• 2 talks (each ca. 40 min incl. discussion)

• Time for networking + food + drinks before, in between and after the presentations

• Talks are held in English

• We will be taking photos and/or film footage at the event. These will be used to share news about our meetups, and to publicize upcoming events.

• We will provide a list of registered usernames to the sponsor to check at the entry.

• Please bring photo identification to be checked at the entrance.

The line up:

Christopher Prohm - Dynamic obstacle avoidance in autonomous racing

Abstract:
In a recent collaboration between Volkswagen Data:Lab, Italdesign Giugiaro, and Roborace, the partners worked on dynamic obstacle avoidance in autonomous racing. Using the autonomous Roborace DevBot as a platform, Volkswagen Data:Lab and Italdesign Giugiaro developed and tested new algorithms for dynamic object detection and path planning. On a high level, three steps are preformed to achieve this goal. First objects are detected via lidars as the car approaches them. Based on this information a safe trajectory around the obstacle is planned. Finally, a controller uses the current state of the car and the planned trajectory to determine appropriate steering and throttle controls to follow the path. This talk will give an overview of the project. Starting from its setup, the talk will cover the different components that were developed within the project and discuss how they are combined. A particular focus will be the application of machine learning to trajectory planning. The talk will also show how development and track testing looked like.

Bio:
Christopher is working as Data Scientist for Volkswagen Data:Lab. Trained as Physicist, he is now focusing on the application of machine learning to engineering topics. Within this application domain, he is working at the intersection of modern machine learning and classical engineering approaches.

Artemii Frolov - Real-time emotion recognition during presentations

Abstract:
Understanding of emotions during presentations could be very important for the presenter. It could be not only useful for him but also make presentations more interesting for the listeners. In this talk, I would like to present my solution, based on Amazon DeepLens, and using MTCNN and mini-Xception networks (with an explanation of how they work). We'll discuss easy visualization of the results and how to expand the solution for every mini-computer with a camera.

Bio:
Artemii is a Master student at Technical University of Munich with a specialization in Data Science and Machine Learning. He has graduated in 2015 with an Engineering degree in Informatics and later worked on research and applied Data Science projects in Accenture, Data Reply and several Russian National Research Centers. His main research interests are Computer Vision and NLP.