AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

RL-ViGen: A Reinforcement Learning Benchmark for Visual Generalization

Zhecheng Yuan, Sizhe Yang, Pu Hua, Can Chang, Kaizhe Hu, Huazhe Xu

NeurIPS 2023 · Reinforcement learning › Visual and representation RL · approach: Robust visual encoders for RL · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use ImageNet pre-trained ResNet encoder (PIE-G) for visual generalizationPIE-G · Habitat ImageNav appearance/lightingsuccess rateImproved
segment agent via saliency to remove background (SGQN)SGQN · Habitat and CARLA (object-rich, first-person)generalization scoreHurt
frequency-domain data augmentation (SRM)SRM · CARLA autonomous drivingaggregated returnImproved
ImageNet pre-trained encoder under low-frequency scenes (dark night)PIE-G · CARLA autonomous drivingaggregated returnHurt
strong augmentation versus random shift only, with VRL3 safe Q mechanismVRL3-based agents · Adroit sample efficiencysuccess rateNo clear effect
ImageNet pre-trained model for fine-grained features (PIE-G)PIE-G · Adroit hard visual generalizationsuccess rateImproved
saliency-map based generalization (SGQN)SGQN · Adroit camera views, hardsuccess rateImproved
generalization methods for scene structure changesall algorithms · CARLA scene structuresreturnNo clear effect
generalization-based methods versus sample-efficiency-focused methodscross-embodiment transfer (Panda to KUKA IIWA, Kinova3) · Robosuite cross embodimentreturnMixed
dynamic video backgrounds in hard settingSRM · Robosuite table-top hardreturnHurt
single shared encoder without target encoder or momentum update in CURLCURL re-implementation · DMControl sample efficiencyepisode returnImproved
add pre-trained encoder inference or saliency mapsPIE-G and SGQN · DMControl Walker walkFPSHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy