AirPlayground: Benchmarking Aerial Navigation across Diverse Reconstructed Environments
To address these limitations, we introduce AirPlayground, a unified benchmark for training and evaluating aerial navigation policies across diverse environments.
ProofPaper ↗
Key points
- Large-scale aerial resources provide a foundation for learning navigation across diverse environments.
- We first construct 94 photorealistic 3D Gaussian Splatting (3DGS) environments spanning indoor, urban outdoor, and vegetation settings, with aligned collision geometry.
- Building on these environments, we develop a scene-adaptive pipeline for automated expert flight collection, yielding 94K demonstrations and 46.5M synchronized observation-state-action samples at 50 Hz.
- We benchmark seven baselines spanning four policy paradigms across three navigation settings in seen and unseen environments, and further examine scaling behavior with ACT.
Sources (1)
- [1]AirPlayground: Benchmarking Aerial Navigation across Diverse Reconstructed EnvironmentsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 21, 01:13 PM
To address these limitations, we introduce AirPlayground, a unified benchmark for training and evaluating aerial navigation policies across diverse environments.
Large-scale aerial resources provide a foundation for learning navigation across diverse environments.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 21, 2026AOV-GS: Active Open-Vocabulary Semantic Mapping with 3D Gaussian Splatting
- Oct 20, 2026KoalaGS: Complexity-Aware Continuous Level of Detail for Gaussian Splatting
- Oct 20, 2026Learning from Automatic 3D Gaussian Splatting Edits for Online Bayesian Scene Change Detection
- Oct 20, 2026StreamEval: Evaluating Interactive Streaming Video Generation Models
- Oct 19, 2026SHAM-Bench: A Span-level Hallucination Benchmark Across Multiple Tasks
- Oct 19, 2026Learning Geometry-Aware Planning Representations for Monocular Flight with Cross-Modal Supervision