REINFORCEMENT-LEARNINGCURRENT2026-05-12

EvoNav: Evolutionary Reward Function Design for Robot Navigation with Large Language Models

Zhikai Zhao, Chuanbo Hua, Federico Berto, Zihan Ma, Kanghoon Lee, Jiachen Li, Jinkyoo Park

Instead of manually tuning Imitation & Reinforcement LearningRewardA score that tells the robot how well it is doing. functions for Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. Navigation & LocomotionNavigationMoving through an environment toward a goal. (which requires domain expertise and often fails), EvoNav uses LLMs to automatically generate and evolve better Imitation & Reinforcement LearningRewardA score that tells the robot how well it is doing. functions. The system validates candidates cheaply with proxies before expensive full Robot LearningTrainingThe process of fitting a model using data or experience., producing Navigation & LocomotionNavigationMoving through an environment toward a goal. policies that outperform hand-crafted rewards.

THE PROBLEM

This paper focuses on Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards.. EvoNav automates Imitation & Reinforcement LearningReward functionThe rule that defines how rewards are assigned. design for Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. Navigation & LocomotionNavigationMoving through an environment toward a goal. in dynamic human environments using evolutionary algorithms and LLMs. It addresses the key Imitation & Reinforcement LearningReinforcement Learning (RL)Teaching a robot through trial and error using rewards. problem of Imitation & Reinforcement LearningRewardA score that tells the robot how well it is doing. specification sensitivity through a three-stage Simulation & Sim-to-RealEvaluationMeasuring how well a robot system performs. pipeline (analytical proxies → lightweight rollouts → full Robot LearningTrainingThe process of fitting a model using data or experience.) that reduces computational cost while improving Core ConceptsPolicyThe rule or model that maps observations or states to actions. quality. 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

Instead of manually tuning Imitation & Reinforcement LearningRewardA score that tells the robot how well it is doing. functions for Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. Navigation & LocomotionNavigationMoving through an environment toward a goal. (which requires domain expertise and often fails), EvoNav uses LLMs to automatically generate and evolve better Imitation & Reinforcement LearningRewardA score that tells the robot how well it is doing. functions. The system validates candidates cheaply with proxies before expensive full Robot LearningTrainingThe process of fitting a model using data or experience., producing Navigation & LocomotionNavigationMoving through an environment toward a goal. policies that outperform hand-crafted rewards. 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

Instead of manually tuning Imitation & Reinforcement LearningRewardA score that tells the robot how well it is doing. functions for Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. Navigation & LocomotionNavigationMoving through an environment toward a goal. (which requires domain expertise and often fails), EvoNav uses LLMs to automatically generate and evolve better Imitation & Reinforcement LearningRewardA score that tells the robot how well it is doing. functions. The system validates candidates cheaply with proxies before expensive full Robot LearningTrainingThe process of fitting a model using data or experience., producing Navigation & LocomotionNavigationMoving through an environment toward a goal. policies that outperform hand-crafted rewards.

WHY DEVELOPERS SHOULD CARE

Instead of manually tuning Imitation & Reinforcement LearningRewardA score that tells the robot how well it is doing. functions for Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. Navigation & LocomotionNavigationMoving through an environment toward a goal. (which requires domain expertise and often fails), EvoNav uses LLMs to automatically generate and evolve better Imitation & Reinforcement LearningRewardA score that tells the robot how well it is doing. functions. The system validates candidates cheaply with proxies before expensive full Robot LearningTrainingThe process of fitting a model using data or experience., producing Navigation & LocomotionNavigationMoving through an environment toward a goal. policies that outperform hand-crafted rewards.

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