SA-GCS: Semantic-Aware Gaussian Curriculum Scheduling for UAV Vision-Language Navigation

TL;DR

This paper introduces SA-GCS, a novel training framework that combines semantic-aware difficulty estimation with curriculum scheduling to improve UAV vision-language navigation, leading to faster convergence and better performance.

Contribution

It proposes a new Semantic-Aware Gaussian Curriculum Scheduling method that enhances reinforcement learning for UAV navigation by systematically adjusting training sample difficulty.

Findings

Outperforms strong baselines on CityNav benchmark

Achieves faster and more stable convergence

Generalizes well across different model scales

Abstract

Unmanned Aerial Vehicle (UAV) Vision-Language Navigation (VLN) aims to enable agents to accurately localize targets and plan flight paths in complex environments based on natural language instructions, with broad applications in intelligent inspection, disaster rescue, and urban monitoring. Recent progress in Vision-Language Models (VLMs) has provided strong semantic understanding for this task, while reinforcement learning (RL) has emerged as a promising post-training strategy to further improve generalization. However, existing RL methods often suffer from inefficient use of training data, slow convergence, and insufficient consideration of the difficulty variation among training samples, which limits further performance improvement. To address these challenges, we propose \textbf{Semantic-Aware Gaussian Curriculum Scheduling (SA-GCS)}, a novel training framework that systematically integrates Curriculum Learning (CL) into RL. SA-GCS employs a Semantic-Aware Difficulty Estimator (SA-DE) to quantify the complexity of training samples and a Gaussian Curriculum Scheduler (GCS) to dynamically adjust the sampling distribution, enabling a smooth progression from easy to challenging tasks. This design significantly improves training efficiency, accelerates convergence, and enhances overall model performance. Extensive experiments on the CityNav benchmark demonstrate that SA-GCS consistently outperforms strong baselines across all metrics, achieves faster and more stable convergence, and generalizes well across models of different scales, highlighting its robustness and scalability. The implementation of our approach is publicly available.