Moving Forward by Moving Backward: Embedding Action Impact over Action Semantics | AI2

This video describes the Action Adaptive Policy by Zeng et al. (ICLR 2023). In the video, we discuss the formulation of motion drifts in point navigation and object navigation tasks. Additionally, we introduce the Action Adaptive Policy (AAP) to address these motion drifts. Finally, we demonstrate that AAP performs well when facing unseen drifts during inference in RoboTHOR, Habitat, PettingZoo, and a real-world RoboTHOR test scene.
------------------
The voiceover for this video was generated from text-to-speech with WellSaid Labs: https://wellsaidlabs.com/

ai2 discuss embedding iclr impact inference moving navigation policy semantics test video world