ForVis: An In-Field Dataset and Benchmark for VIO Using Under-Canopy UAV Flights in Forests

2026-09-28

AI Summary

This study addresses the lack of evaluation benchmarks for UAV visual-inertial SLAM in complex forest environments by constructing the first real-world VIO dataset encompassing multiple scenarios, including open areas and regions above and below the forest canopy. By synchronously collecting camera and IMU data from Intel RealSense D435i and OAK-D Pro Wide sensors, extensive benchmarking was conducted across various VI-SLAM systems. Results from 504 experiments demonstrate that sensor selection critically impacts localization accuracy, with the OAK-D Pro significantly outperforming the D435i in trajectory error. This research provides a vital evaluation benchmark and practical hardware selection guidance for UAV navigation in under-canopy forest environments.

Abstract

Visual-inertial Simultaneous Localization and Mapping (VI-SLAM) for UAVs remains difficult to evaluate in real forest environments, where motion, illumination changes, repetitive vegetation, and vibration can all affect estimation. We present ForVis, an in-field dataset and benchmark for evaluating VI-SLAM during UAV flight in forest environments. The dataset contains twelve flights across open meadow, above-canopy, and under-canopy conditions in each environment. In total, it provides 563.8s of flight over 1096.8m of trajectory, recorded simultaneously with an Intel RealSense D435i and an OAK-D Pro Wide together with inertial and flight-controller data. We benchmark seven open-source VI-SLAM systems over 504 runs. The results show that sensor choice has a larger effect on trajectory error than the spread between algorithms: all seven methods achieve lower median error on the OAK-D Pro than on the D435i. ForVis is intended to support evaluation of speed, accuracy and robustness for VI-SLAM in challenging forest flight.

Problem

• VI-SLAM
• UAV
• forest environment
• dataset
• benchmark

Innovation

• Visual-Inertial SLAM
• UAV
• Forest Dataset
• Benchmark
• Sensor Evaluation