Master Thesis Multi-Modal Active User Detection and Command Identification for Service Robots

Service robots in diverse human environments need to recognize and understand the movements as well as activities of the surrounding people to enable socially-compliant navigation and operation. To that end, robots are equipped with human detection and tracking, full-body pose and gaze estimation, speech and gesture recognition, activity classification and future motion prediction. In addition to that, providing active services and receiving commands requires recognizing user engagement from the passive third-party bystanders. In this thesis we intend to develop a robust pipeline for active user detection, engagement level estimation and robot response generation using the multi-modal robot perception input. During your master thesis, you will research and implement novel algorithms for human-robot interaction with service robots, which operate in social and industrial spaces. Base your work on state-of-the-art robot perception techniques. Furthermore, you will contribute to improved human-awareness, safety and efficiency of service, domestic and intralogistics robots. Moreover, you will be responsible for the design, implementation, verification, and test of the algorithms. Additionally, you will support the integration and test of your algorithms in simulation and on a real robot. Last but not least, you will work in an agile and diverse research team and collaborate deeply with experts in the field of human-oriented robotics. Beyond that you benefit from powerful infrastructure to accelerate your research.