SNeRL: Semantic-aware Neural Radiance Fields for Reinforcement Learning

* Status: accepted for publication in the Conference on Machine Learning (ICML) 2023
* Category: Autonomous Robot Learning
* Author : Dongseok Shim*, Seungjae Lee*, and H. Jin. Kim (* Equal contribution)
* Abstract: As previous representations for reinforcement learning cannot effectively incorporate a human-intuitive understanding of the 3D environment, they usually suffer from sub-optimal performances. In this paper, we present Semantic-aware Neural Radiance Fields for Reinforcement Learning (SNeRL), which jointly optimizes semantic-aware neural radiance fields (NeRF) with a convolutional encoder to learn 3D-aware neural implicit representation from multi-view images. We introduce 3D semantic and distilled feature fields in parallel to the RGB radiance fields in NeRF to learn semantic and object-centric representation for reinforcement learning. SNeRL outperforms not only previous pixel-based representations but also recent 3D-aware representations both in model-free and model-based reinforcement learning.
* Contact : [e-mail], [e-mail]