REVIEW 9 cited by
The Distracting Control Suite -- A Challenging Benchmark for Reinforcement Learning from Pixels
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Robots have to face challenging perceptual settings, including changes in viewpoint, lighting, and background. Current simulated reinforcement learning (RL) benchmarks such as DM Control provide visual input without such complexity, which limits the transfer of well-performing methods to the real world. In this paper, we extend DM Control with three kinds of visual distractions (variations in background, color, and camera pose) to produce a new challenging benchmark for vision-based control, and we analyze state of the art RL algorithms in these settings. Our experiments show that current RL methods for vision-based control perform poorly under distractions, and that their performance decreases with increasing distraction complexity, showing that new methods are needed to cope with the visual complexities of the real world. We also find that combinations of multiple distraction types are more difficult than a mere combination of their individual effects.
Forward citations
Cited by 9 Pith papers
-
Latent Actions from Factorized Transition Effects under Agent Ambiguity
OTF decomposes transitions into reusable primitives to form action-like latents in OTF-LAM and OTF-LAM-Dino, enabling zeroshot transfer and competitive policy learning under visual ambiguity.
-
Agent-Centric Observation Adaptation for Robust Visual Control under Dynamic Perturbations
ACO-MoE employs agent-centric mixture-of-experts to decouple task-relevant features from dynamic visual perturbations in RL, recovering 95.3% of clean performance on the new VDCS benchmark.
-
Agent-Centric Observation Adaptation for Robust Visual Control under Dynamic Perturbations
ACO-MoE recovers 95.3% of clean-input performance in visual control tasks under Markov-switching corruptions by routing restoration experts and anchoring representations to clean foreground masks.
-
Margin in Abstract Spaces
Sufficiently large margins make distance-based concept classes learnable in every metric space via the triangle inequality alone, with a sharp universal threshold and a negative embedding result.
-
Latent Actions from Factorized Transition Effects under Agent Ambiguity
Factorizing pixel transitions into a learned codebook of patch-level motion primitives, then gating them into latent actions, transfers across morphologies and matches or beats monolithic latent-action baselines in di...
-
Why Latent Actions Fail, and How to Prevent It
Extending linear LAMs to model exogenous state shows standard reconstruction encodes future exogenous info in latent actions, while endogenous-focused spaces and auxiliary objectives like action-supervision enforce co...
-
ELVIS: Ensemble-Calibrated Latent Imagination for Long-Horizon Visual MPC
ELVIS achieves state-of-the-art results on 14 visual control tasks and zero-shot real-world transfer by using ensemble-calibrated lambda-returns and Gaussian-mixture MPPI inside a latent RSSM planner.
-
LeapBot-WA: World-Anchor Action Models via Predictive Latent Alignments
LeapBot-WA shows robot policies can be trained with latent world-model predictions instead of pixel video generation, hitting state-of-the-art for predictive action models and staying competitive with generative WAMs.
-
stable-worldmodel: A Platform for Reproducible World Modeling Research and Evaluation
The paper presents stable-worldmodel (swm), a platform with high-performance data layer, modern world model baselines, planning solvers, and extended environments for reproducible research and generalization evaluation.
Discussion (0). Sign in to comment.