On December 5, we heard Robert Ross talk about „Squeezing Shared Meaning from Behaviour“. Thank you for being here!
Dr. Robert Ross is a Senior Lecturer in the School of Computing DIT. Robert has a BSc and MSc by Research from University College Dublin, and a PhD from the University of Bremen in Germany. Robert is responsible for managing mature learning programmes within the School.
Abstract of the talk: To understand and facilitate a human is to provide a clear understanding of a persons’s activities or behaviours and subsequently provide feedback and personalised services. While Activity Recognition based on well defined activity prototypes is now well established, the related task of Activity Discovery – the extraction of new or changed behaviours from the data – remains highly challenging. In this talk I will discuss recent research conducted in Dublin where we apply a number of different methodologies including Deep Learning techniques to extract new behaviour prototypes from low-level sensor data. I will also discuss some challenges around the evaluation of the quality of extracted behaviours, and raise some questions around the goal of linking behaviour prototypes of linguistic meaning.
A TV team from RTL Nord was visiting our lab yesterday to check out our robots as well as our Virtual Reality kitchen environment.
The workshops on Latest Advances in Big Activity Data Sources for Robotics & New Challenges at IROS 2018 were a great success!
Fall School has started, here are the first impressions!
EASE principal investigators Alin Albu-Schäffer, Michael Beetz, and Helge Ritter gave invited talks at the French-German Symposium on “Robotics AI – Data Science versus Motion Intelligence” co-organized by the French Académie des Sciences and the German Nationale Akademie der Wissenschaften Leopoldina. https://robotics-ia.sciencesconf.org/
On Thursday, 29 June 2018, we had a successful open lab day and had the opportunity to welcome many interested visitors to our lab. We would like to thank everybody for interesting discussions, new ideas and a useful exchange about opinions about the future of robotics. We would also like to thank the DFKI Bremen for the cooperation.
The EASE milestone meeting turned out to be a great success on all levels. Our milestones are met, if not exceeded and our project is moving forward in all directions. We were able to see how to collect big data from human table setting scenarios, the enhancements in the Virtual Reality environment Unreal – including a better grasp – as well as the new software Avatar, which now includes humans in our virtual environments, just to name a few.
Some impressions from the meeting:
EASE at the Automatica collected data from random participants in virtual reality scenarios. The more data we collect, the more variations on the executed tasks we can analyze!
Three tasks were investigated:
– doing the dishes (Kitchen)
– setting the table (Dinning room)
– vacuum cleaning (Living room)
Thank you to more than 200 participants!