Personal Robot Coach
This project explores how the embodiment and feedback of a social robots as well as the personalization of a body weight workout affect the user's motivation to do sports.
Diseases of affluence, like obesity, are one of the major challenges in the 21th century. People tend to eat unhealthy and do less sportive activities. Governments try to counter these developments with funded projects. However, those approaches do not achieve success yet. People lack of time or enthusiasm to exercise regularly. This opens the focus for new approaches to engage and motivate people to work out every day with enthusiasm and goal orientation. Technological advances in the consumer market like smartphone applications or exergaming (Nintendo Wii, Microsoft Kinect) show great potential for changing exercising habits. After all, a limiting factor of current studies is the novelty effect.
Therefore, research is needed which systematically investigates the different factors that shape and establish long-term relationships between robots and humans. In this work we focus on the aspects of feedback, adaption and preference learning for socially assistive robots (SAR) that are important for long-term interaction. To examine these after novelty effects wear off we create an application scenario where users are accompanied by a social robot to do a repeated exercising workout. The main goal for the adpation process is to find the prefered exercises and workout intensity for each participant during interaction. Thus, this research project will intersect with three main topics in HRI: socially assistance, long-term interaction and machine learning techniques for adaption and personalization.
Recent Sport Scenarios
Rowing Body Weight Workout
The central research questions of this project is how and if a social robot can boost the user's motivation to exercise in the long-term. Therefore several secondary issues have to be investigated:
- RQ1: How does the companion style of a robot effects the motivation .?
- RQ2: What effects does the embodiment have on the impression of the robot ?
- RQ3: Can motivational feedback provided by a robot help users to exercise longer ?
- RQ4: How can a system adapt to the user's preferences in online interactions?
Current Project Results
We have investigated whether the Koehler Effect can be replicated with humanoid companions. We found that people experience a motivational gain when exercising with a robotic companion (RC )compared to working out individually (IC) or with a robot instructor (RI) .
We found that acknowledging feedback from a robotic instructor (RIF) during body weight training enhances the user's motivation to a robot instructor that does not give feedback (RI) .
During an 18-days long-term isolation study we compared a robotic instructor versus a computer display for indoor cycling exercising. We found that participants had higher exercise compliance in the robot condition. .
Preference Learning and Adaptationin HRI:
We found that the embodidment of the learning agent does not influence the user's perception of the adaptation quality. Furthermore, we showed that a dueling bandit learning approach is suitable for HRI and achieves better preference results than a randomizied learning approach .
2016 | Conference Paper | PUB-ID: 2906151Exercising with a Humanoid Companion is More Effective than Exercising AlonePUB | PDF
Schneider S, Kummert F (2016)
In: Proceedings of the IEEE-RAS Conference on Humanoid Robots.
2017 | Conference Paper | PUB-ID: 2909140Does the User's Evaluation of a Socially Assistive Robot Change Based on Presence and Companionship Type?PUB | DOI
Schneider S, Kummert F (2017)
In: Proceedings of the Companion of the 2017 ACM/IEEE International Conference on Human-Robot Interaction - HRI '17. Association for Computing Machinery (ACM).
2016 | Conference Paper | PUB-ID: 2906152Motivational Effects of Acknowledging Feedback from a Socially Assistive RobotPUB | PDF
Schneider S, Kummert F (2016)
In: Proceedings of the Eight International Conference on Social Robotics. Springer.
2017 | Conference Paper | PUB-ID: 2909463Dueling Bandit Learning for Preference Adaptation in Human-Robot Interaction*PUB
Schneider S, Kummert F (Submitted)
In: Proceedings of the 26th IEEE International Symposium on Robot and Human Interactive Communication.
2015 | Conference Paper | PUB-ID: 2787035Long-Term Feedback Mechanisms for Robotic Assisted Indoor Cycling TrainingPUB | PDF
Schneider S, Süssenbach L, Berger I, Kummert F (2015)
In: Proceedings International Conference on Human-Agent Interaction. ACM: 157.
Recent Best Paper/Poster Awards
Philippsen A, Reinhart F, Wrede B (2016)
International Conference on Development and Learning and on Epigenetic Robotics (ICDL-EpiRob)
Richter V, Carlmeyer B, Lier F, Meyer zu Borgsen S, Kummert F, Wachsmuth S, Wrede B (2016)
International Conference on Human-agent Interaction (HAI)
Carlmeyer B, Schlangen D, Wrede B (2016)
International Conference on Human Agent Interaction (HAI)