REINFORCEMENT-LEARNINGCURRENT2026-04-27

Agent-Centric Visual Reinforcement Learning under Dynamic Perturbations

Zhengru Fang, Yu Guo, Fei Liu, Yuang Zhang, Yihang Tao, Senkang Hu, Wenbo Ding, Yuguang Fang

Visual Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. agents typically fail when camera feeds degrade (blur, Data, Distributions & Training IssuesNoiseUnwanted variation or randomness in sensor readings or actuation., lighting shifts) because they try to reconstruct corrupted images perfectly. This paper shows you can recover 95%+ performance by routing corrupted inputs through specialized restoration experts that extract only task-relevant features before the Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. agent sees them.

THE PROBLEM

This paper focuses on Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards.. Introduces Visual Degraded Control & PlanningControlThe method used to make the robot move the way you want. Suite (VDCS) Simulation & Sim-to-RealBenchmarkA standard test used to compare methods fairly. extending DMControl with non-stationary visual corruptions. Proves via information theory that reconstruction-based visual Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. fails under perturbations. Proposes ACO-MoE (Agent-Centric Observations with Mixture-of-Experts) using multiple specialized restoration experts to decouple Perception & SensingPerceptionThe process of turning raw sensor data into useful understanding of the world. from perturbation artifacts. Read the paper by tracking the Core ConceptsTaskThe job the robot is supposed to complete, such as pick-and-place, navigation, or drawer opening. definition, the Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. or data assumptions, and the evidence that supports the claimed improvement.

HOW IT WORKS

1

Task framing

The paper frames the work as Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards.. Start here because it defines what success means and which assumptions the rest of the method inherits.

2

Core method

Visual Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. agents typically fail when camera feeds degrade (blur, Data, Distributions & Training IssuesNoiseUnwanted variation or randomness in sensor readings or actuation., lighting shifts) because they try to reconstruct corrupted images perfectly. This paper shows you can recover 95%+ performance by routing corrupted inputs through specialized restoration experts that extract only task-relevant features before the Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. agent sees them. When reading the method section, identify the inputs, the learned or engineered representation, and the Core ConceptsActionA command the robot sends to its motors, controller, or low-level system. or prediction produced by the system.

3

Data and supervision

For robotics work, the data story is part of the method: check whether the system depends on Imitation & Reinforcement LearningTeleoperation (teleop)A human remotely controlling the robot, often to collect demonstrations., Simulation & Sim-to-RealSimulationA virtual environment where robots can be trained or tested., internet video, human labels, or Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. rollouts.

4

Evaluation evidence

The paper should be judged through its Simulation & Sim-to-RealEvaluationMeasuring how well a robot system performs. protocol: what data is used, what Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. or simulator is tested, and which Evaluation & ResearchBaselineA reference method used for comparison. comparisons support the claim. Look for the gap between the headline result and the Simulation & Sim-to-RealDeploymentPutting the trained system on a real robot. setting you would actually care about.

KEY RESULTS

Main contributionConceptual contribution

Visual Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. agents typically fail when camera feeds degrade (blur, Data, Distributions & Training IssuesNoiseUnwanted variation or randomness in sensor readings or actuation., lighting shifts) because they try to reconstruct corrupted images perfectly. This paper shows you can recover 95%+ performance by routing corrupted inputs through specialized restoration experts that extract only task-relevant features before the Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. agent sees them.

WHY DEVELOPERS SHOULD CARE

Visual Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. agents typically fail when camera feeds degrade (blur, Data, Distributions & Training IssuesNoiseUnwanted variation or randomness in sensor readings or actuation., lighting shifts) because they try to reconstruct corrupted images perfectly. This paper shows you can recover 95%+ performance by routing corrupted inputs through specialized restoration experts that extract only task-relevant features before the Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. agent sees them.

LIMITATIONS

The main limitation to check is whether the claimed behavior holds outside the paper's reported setup. That means testing across different Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. embodiments, scenes, objects, and data distributions.

WHAT COMES NEXT

The practical next step is independent reproduction with clear baselines, ablations, and stress tests. For a developer, the useful follow-up is to map the paper's Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. assumptions onto a concrete Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. stack, then test the smallest version of the method that could run end to end.

RELATED PAPERS