HERCULES: Multi-Robot Photorealistic Synthetic SLAM Dataset

HERCULES is a photorealistic synthetic dataset for multi-robot Simultaneous Localization and Mapping and collaborative perception. It was created by LunarLab at Georgia Tech and features data from 2 aerial and 2 ground robots across 4 large-scale environments, including Australian Outback, City Block, and Forest scenes. The dataset page was last updated on 2026-06-29.

Use Cases

• Developing multi-robot SLAM algorithms based on data from 2 aerial and 2 ground robots.
• Training collaborative perception models based on photorealistic synthetic data from four distinct environments.
• Benchmarking autonomous navigation systems in simulated large-scale environments like the Australian Outback and City Block.

Strengths

• Covers 4 distinct large-scale photorealistic environments.
• Includes data from 4 robots (2 aerial, 2 ground), enabling multi-agent scenarios.
• Created by a research lab (LunarLab @ Georgia Tech) with an associated arXiv paper.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and data size are unknown, which may limit suitability assessment.
• The dataset is synthetic, which may not perfectly reflect real-world sensor noise and dynamics.

Provenance

Source

LunarLab @ Georgia Tech

Collection Method

Synthetic data generation in photorealistic simulated environments.

Freshness

Last updated 2026-06-29 18:09:38; freshness should be verified.

Geography

Simulated environments include Australian Outback, City Block, and Forest.