Cognitive Science
Our research on robot learning is also connected with the study of the human mind and modeling of human behavior. By increasing our understanding of the way humans learn to perform novel tasks, we can discover novel ways for robot learning. Conversely, insights from robot learning may increase our understanding of the human mind. For this reason, our group is carrying out research in the following areas: Models of Locomotion, Modeling Human Behavior, Human-Robot Interaction.
Models of Locomotion
Dealing with whole body movement of humanoid robot is a highly challenging domain involving various different problems ranging from stabilizing behaviors (e.g., balancing, support oneself against objects) over static motions (e.g., getting up from a chair, push-ups) up to locomotion (e.g., walking, running). As of now, no robot exhibits the same dexterity, efficiency, speed and robustness as a human due to the difficulty of controlling and planning with a high number of degrees of freedom as well as due to the complexity of modeling and estimating physical contacts. We aim to develop new approaches to improve the current state-of-the-art for whole body movement of humanoid robots in general, and, especially, for robot locomotion.
TODO: Rewrite text to be about human locomotion and not humanoid robotics.
TODO: What references to put here?
Modeling Human Behavior
Insights from control engineering and statistics improve our understanding of how humans work and vice versa. The notions of feedback and information became deeply ingrained in social sciences, following the pioneering work of Norbert Wiener on cybernetics. Creating models of human behavior based on statistics and control theory that capture certain aspects of cognition and intelligence helps in answering fundamental questions about the subconscious processes taking place in a human being. How do we decide which muscles to activate to move an arm? What makes eyes move to track a moving target? Many such questions can be addressed by means of simplified mathematical models. In our research, we develop probabilistic models and inference techniques to gain a better understanding of the amazing human learning abilities. Our models reproduce characteristic features like motor variability, continuous exploration, stochastic decisions and the ability to learn task abstractions.
-
- Tanneberg, D.; Rueckert, E.; Peters, J. (2020). Evolutionary Training and Abstraction Yields Algorithmic Generalization of Neural Computers, Nature Machine Intelligence, 2, 12, pp.753-763.
-
- Belousov, B.; Neumann, G.; Rothkopf, C.; Peters, J. (2016). Catching Heuristics Are Optimal Control Policies, Advances in Neural Information Processing Systems (NIPS / NeurIPS).
Human-Robot Interaction
Due to the inherent stochasticity and diversity of human behavior, robots need learning and adaptation capabilities to successfully interact with humans in various scenarios. Predicting human intent and modeling human behavior is essential for seamless human-robot interaction. Fundamental research in human-robot interaction has potential applications in a variety of scenarios where humans need assistance: assembly of products in factories, the aid of the elderly at home, control of actuated prosthetics, shared control in repetitive teleoperated processes, interaction with home robots and humanoid robot assistants. We have developed machine learning algorithms that enable robots to learn interactions from demonstrations. Moreover, our group has investigated how robots can support the practice and execution of movements by humans. Applications of our methods have been demonstrated in tasks involving humans and robots working in partnership, in teleoperation scenarios with shared autonomy between the human and the robot, and in interaction of humans with humanoid robot assistants, among other situations.
-
- Li, Q.; Chalvatzaki, G.; Peters, J.; Wang, Y. (2021). Directed Acyclic Graph Neural Network for Human Motion Prediction, Proceedings of the IEEE International Conference on Robotics and Automation (ICRA).