Informatik-Oberseminar: Uncertainty-Aware Perceptual Robot Learning

Donnerstag, 27.06.2019, 14.15 Uhr

Ort: UMIC Research Centre, Raum 025, Mies-van-der-Rohe-Straße 15

Referent: Prof. Dr. David Held, The Robotics Institute, Carnegie Mellon University

Abstract:

Robot learning systems need to deal with perceptual uncertainty, due to a variety of factors such as occlusions, sensor noise, small objects, limited capacity models, and novel objects.  We believe that perceptual systems should be aware of their uncertainty and that decision-making algorithms should incorporate such uncertainty.

We present three current directions for achieving this.  

First, we present an object instance detection system that combines machine learning with correspondence matching to verify the proposed detections and reject uncertain detections.  

Next, we show an approach for estimate a distribution over orientation uncertainty by augmenting any deep pose estimation system.  

Last, we present an approach for reinforcement learning with deformable objects that does not assume access to the ground-truth state.  

Together, these directions represent our current approach to dealing with perceptual uncertainty in robot learning systems.

 

Es laden ein: die Dozentinnen und Dozenten der Informatik