This website introduces a data set comprised of robot sensor, ground truth and system-level data which we have recorded at the end of year 2 of the HUMAVIPS project. In this data set we addressed the issues named below by choosing a natural interaction scenario inspired from conversational analysis and by recording data in a whole-system approach including many internal of the robot NAO, the target platform of the project.
The HUMAVIPS project aims to endow the humanoid robot NAO with audio-visual abilities to manage the interaction with a group of several people. In order to develop these abilities, computer scientists require a corpus of relevant data which allows to
- a) understand the behavior of people interacting with a humanoid robot to create appropriate models of the interaction and algorithms to analyze the current situation
- b) benchmark the performance of developed algorithms
- c) verify the interplay of components in the whole robotic system.
Data Set Access
The data set is available to the research community on request.
Please use the contact page to request access.
The main features are
- human robot interaction with multiple participants based on established robot platform NAO
- focus on whole system perspective including system-internal data
- synchronized data
- ground truth information