The virtual agent Flobi is a representation of the humanoid robot head Flobi. This simulation platform allows us to conduct various scenarios of human-robot interaction. The simulation has been developed using the Unity3D Game Engine [1] in conjunction with the Unity-Robotics-Hub [2]. By using ROS as the interface for robot simulation, functions can be developed that can be used for both simulated and real robots, promoting reusability. Our ongoing efforts are focused on improving the processing of sensory inputs and adaptive responses for Flobi to enable a seamless and natural dialogue between humans and the agent.
With the CSRA, a complex supportive environment has been developed by combining a smart home with a cognitive social robot whose advanced manipulation capabilities enable it to distinguish between different social situations that require different social behaviour. This enables helpful and unobtrusive 24/7 operation. Furthermore, this approach will open up a new level of interactive capabilities, raising new research questions about the social and technical aspects of long-term human-technology interactions.
The Collaborative Research Centre/Transregio "Constructing Explainability" (TRR 318), newly established by the DFG at the Universities of Bielefeld and Paderborn as of 1 July 2021, deals with the question of how transparency of algorithmic decisions can be established, especially through black-box methods of modern artificial intelligence. The central hypothesis of TRR 318 is that explanations are most effective when they are co-constructed by both the explainer and the receiver.
Birte Richter's (doctoral) thesis "The Attention-Hesitation Model. A Non-Intrusive Intervention Strategy for Incremental Smart Home Dialogue Management" deals with so-called Smart Personal Assistants (SPAs), which are finding their way into the household and support us in our daily activities. Currently, these agents do not respond to the attention of the smart home user. However, we know from research on human-human interaction (HHI) that people coordinate their speech and continuously adapt their behaviour to the actions and reactions of their interaction partner. The central question Birte Richter poses in her (doctoral) thesis is therefore how human attention can be incorporated into dialogue management to improve human-agent interaction (HAI) in smart homes.
The ADHD attention training with the virtual agent Flobi (RoboCamp) is tailored to patients with ADHD. They have a deficit due to delayed reward processing to learn positive or negative consequences (delay aversion due to delayed reward). Therefore, they need a prompt immediate reward or negative consequence. In human-human interaction, an intensive, effective behavioural therapy training programme for children and adolescents with ADHD (the Summer Camp) has been shown to have lasting effects. The main feature of the training is an intensive response cost intervention (RCI).
In project A03, the researchers include the aspect of emotion in the explanation process: How do emotions influence the perception of an AI explanation? What decisions do people make with the help of artificial intelligence in risk situations? How can explanations be adapted to the emotional state of humans? The computer scientists and economists are investigating how emotions such as joy and fear affect understanding processes and decisions. Their goal is to develop a model for an intelligent machine that recognises emotions and takes them into account in explanation processes.
Project A05 deals with human-robot interaction and analyses how attention can be directed in the explanation process. The aim of the experimental investigations is to understand whether certain types of attention guidance at the micro and task level of the interaction lead to different characteristics of the attention parameters in humans. These can then be implemented in the behaviour of a robot in terms of scaffolding and monitoring. The investigations on the task level deal with the history of the interaction, which is a prerequisite for the emergence of a social practice.
ADHD is often overlooked or misdiagnosed in people with intellectual disabilities (ID) - with far-reaching consequences for everyday life and quality of life. The GAZE-AID research project is therefore developing an innovative, technology-supported diagnostic approach based on eye-tracking. The aim is to develop a reliable, objective and accessible method that is specifically tailored to the needs of people with IB. The aim is to reduce misdiagnosis and overmedication and enable targeted treatments.
IDWorkSupport is an innovative assistance system that supports people with intellectual disabilities (ID) with attention training in everyday working life.
The system is designed to help improve attention, concentration and cognitive skills to enable better participation in working life.