From Robot Learning To Robot Understanding: Leveraging Causal Graphical Models For Robotics

Causal graphical models have been proposed as a way to efficiently 1 and explicitly reason about novel situations and the likely outcomes of decisions. 2 A key challenge facing widespread implementation of these models in robots is 3 using prior knowledge to hypothesize good candidate causal structures when the 4 relevant environmental features are not known in advance. The tight link between 5 causal reasoning and the ability to intervene in the world suggests that robotics 6 has much to contribute to this challenge and would reap significant benefits from 7 progress. 8