Navigator FCN Model Hyperparameters for Vision-Based UAV Navigation

A hyperparameter dataset for a Fully Connected Network (FCN) model used in autonomous UAV navigation research. The model integrates depth maps from the DepthAnything-V2 estimator with Vision-Language Model (VLM) outputs and relative heading angles to predict navigation actions. The dataset, 5.5 KB in size, was authored by Gobinda Chandra Sarker and last updated on April 3, 2026.

Use Cases

• Replicate or fine-tune the Navigator FCN model based on the described hyperparameters for UAV navigation.
• Benchmark hyperparameter configurations for models integrating depth estimation and Vision-Language Model outputs.
• Study the relationship between model hyperparameters and the reported 0.98 task completion rate in AirSim simulations.

Strengths

• Dataset is small (5.5 KB), facilitating quick download and inspection.
• Associated research describes a specific, validated system achieving a 0.98 task completion rate in simulation.
• Released under a permissive CC-BY-4.0 license.

Limitations

• Row count and specific column definitions are unknown, limiting suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data is derived from a specific research simulation; generalizability to other platforms or real-world deployment is untested.

Provenance

Source

Gobinda Chandra Sarker via figshare.

Collection Method

Likely generated as part of research on a vision-based UAV navigation system integrating DepthAnything-V2 and VLMs.

Time Range

Associated research simulations were likely conducted prior to the 2026 dataset update.

Freshness

Last updated 2026-04-03 17:54:42.

Geography

Simulations were conducted in the AirSim Blocks and Downtown West environments; geographic scope of applicability is not specified.