Visual Teach-and-Repeat Navigation Data for Low-Light and Night-Time Environments

A 23.5 GB dataset for visual teach-and-repeat (VTR) navigation designed to operate robustly in environments with variable or low light levels. The data, authored by Fuhai Ling and last updated in May 2026, supports a framework integrating deep-learned descriptors, stereo imaging, and event-based cameras. Experiments demonstrate the system's performance in night-time urban environments for both indoor and outdoor navigation.

Use Cases

• Training and evaluating visual navigation models based on deep-learned local scene descriptors.
• Developing drift error correction algorithms for robotic odometry based on monocular image sequences.
• Benchmarking the performance of event-based cameras against conventional RGB cameras for navigation in low-light conditions.
• Researching topological mapping and decision-making strategies to reduce latency and trajectory error in autonomous systems.

Strengths

• Large-scale dataset at 23.5 GB, suitable for training complex vision models.
• Explicitly designed and tested for robustness in challenging low-light and night-time environments.
• Includes data from both conventional and event-based camera systems, allowing for comparative analysis.
• Released under a permissive CC-BY-4.0 license for open use and modification.

Limitations

• Column-level documentation and sample data are unavailable, requiring manual inspection after download.
• The exact number of rows or data samples is unknown, which may complicate suitability assessment for specific tasks.
• Data collection methodology and specific geographic or environmental details beyond 'night-time urban environment' are not fully specified.

Provenance

Source

Fuhai Ling via figshare.

Collection Method

Likely collected through robotic navigation experiments using stereo and event-based cameras in indoor, outdoor, and night-time urban environments.

Freshness

Last updated 2026-05-20 21:40:39.

Geography

Experiments were conducted in night-time urban environments; specific location is unknown.