Speaker: PhD. Student Emmanuel Senft

Abstract: Robots interacting with humans are expected to behave appropriately according to the current context and partners. As such, it seems highly unlikely that a perfect behaviour suitable for every possible aspects of the interaction can be known in advance by the developers. For this reason, we argue that humans should be able to teach a robot how to interact with other humans. Furthermore, by having a human supervisor in control of the robot actions, we can ensure that robots will always exhibit an appropriate behaviour. This talk will present the concepts of supervised autonomy and interactive machine, and how combining them can allow humans to teach robots how to interact in human environment.