What does this research mean for the field?

Integrating Digital Surface Models (DSM) for metric scaling, Kalman filtering, and data-driven velocity correction into optical flow-assisted UAV navigation reduces velocity and position errors by up to 88% over heterogeneous terrain. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

Mitigate optical flow velocity errors and positioning drift caused by elevation variations and scene texture bias during GPS-denied UAV navigation.

August 29, 2026SensorsOpen Access

Improved Fusion of Optical Flow and Dead Reckoning for UAV Navigation Using Digital Terrain Models and Data-Driven Velocity Correction

View Full Paper

Ask AI

Bookmark

Authors

JWJakub WalczakMilitary University of Technology in Warsaw PTPiotr TargowskiCity University of New York SCSzymon ChmielewskiUniversity of Life Sciences in Lublin

Discussion

Member takes

Implication

Experimental evaluation demonstrates improved velocity accuracy in GPS-denied drone navigation, indicating reduced drift across complex terrain transitions.

Key Points

Mitigate optical flow velocity errors and positioning drift caused by elevation variations and scene texture bias during GPS-denied UAV navigation.
Replaced single-point altitude estimation with camera footprint area-mean sampling using co-registered Digital Terrain Models (DTM) and Digital Surface Models (DSM).
Applied a one-dimensional Kalman filter with an innovation gate to filter out velocity spikes at abrupt terrain transitions.
Trained a data-driven velocity correction factor using flow, texture, and terrain descriptors across 7.3 minutes of PX4 flight log data (Log 258 for calibration; Logs 259–260 for evaluation).
Decreased velocity mean absolute error by 71% (1.04 m/s to 0.30 m/s) on a mixed forest–field validation flight (Log 260).
Reduced dead-reckoning position mean absolute error by 88% (59.8 m to 7.1 m) on the forest–field validation route.
Maintained baseline performance without degradation on a flat open-terrain flight (Log 259), confirming stability over uniform surfaces.

Cite This Study

Walczak et al. (2026) studied this question.

synapsesocial.com/papers/6a9299c88e5d7d1fc0c11ef9 https://doi.org/10.3390/s26175421

View Full Paper

Ask AI

Bookmark

Implication

Experimental evaluation demonstrates improved velocity accuracy in GPS-denied drone navigation, indicating reduced drift across complex terrain transitions.

Improved Fusion of Optical Flow and Dead Reckoning for UAV Navigation Using Digital Terrain Models and Data-Driven Velocity Correction

Discussion

Member takes

Implication

Key Points

Cite This Study

Improved Fusion of Optical Flow and Dead Reckoning for UAV Navigation Using Digital Terrain Models and Data-Driven Velocity Correction

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider