Karin Anna Hummel

Manuela Pollak, Andrea Salfinger, Karin Anna Hummel

We investigate whether naturalistic emotional human feedback can be directly exploited as a reward signal for training artificial agents via interactive human-in-the-loop reinforcement learning. To answer this question, we devise an experimental setting inspired by animal training, in which human test subjects interactively teach an emulated drone agent their desired command-action-mapping by providing emotional feedback on the drone's action selections. We present a first empirical proof-of-concept study and analysis confirming that human facial emotion expression can be directly exploited as reward signal in such interactive learning settings. Thereby, we contribute empirical findings towards more naturalistic and intuitive forms of reinforcement learning especially designed for non-expert users.

Bernhard Etzlinger, Barbara Nußbaummüller, Philipp Peterseil et al.

Mobile contact tracing apps are -- in principle -- a perfect aid to condemn the human-to-human spread of an infectious disease such as COVID-19 due to the wide use of smartphones worldwide. Yet, the unknown accuracy of contact estimation by wireless technologies hinders the broader use. We address this challenge by conducting a measurement study with a custom testbed to show the capabilities and limitations of Bluetooth Low Energy (BLE) in different scenarios. Distance estimation is based on interpreting the signal pathloss with a basic linear and a logarithmic model. Further, we compare our results with accurate ultra-wideband (UWB) distance measurements. While the results indicate that distance estimation by BLE is not accurate enough, a contact detector can detect contacts below 2.5 m with a true positive rate of 0.65 for the logarithmic and of 0.54 for the linear model. Further, the measurements reveal that multi-path signal propagation reduces the effect of body shielding and thus increases detection accuracy in indoor scenarios.