Document Actions

You are here: Home / The Network / Network Partners / Bernstein Focus: Learning / Learning behavioral models

Learning behavioral models

Bernstein Focus: Neuronal Basis of Learning (BFNL)

"Learning behavioral models: From human experiment to technical assistance"

Learning is a pervasive principle across many time-scales and levels of the nervous system. We will focus on an exemplary problem, how humans and robots generate goal-directed actions while situated in natural environments. Learning will be addressed at three time scales: (1) Scene and object memories will be built from single views, and action sequences will be learned from single demonstrations. (2) Online learning of behavioral models of goal-directed movements will occur across many sample experiences. (3) Models of the self-organization of neuronal movement representations will reflect developmental processes.

Human behavioral experiments will provide empirical grounding to process models of learning at all three time scales. In fact, modeling the behavior in these experiments will be the primary tool of data analysis. We will use the models and learning processes to create a motion planning system for Service Robots, enabling these to generate and adapt human-like goal-directed action. Driver models learned on the basis of driving experiments will serve to design, simulate, and test Advanced Driver Assistance Systems.

 


The BFNL 'Learning behavioral models' is a cooperation of scientists at the following research institutions:

  • Ruhr-Universität Bochum

Industry partners:

  • NISYS GmbH, Bochum
  • Schunk GmbH & Co. KG, Lauffen/Neckar