Hybrid Forest Mapping
GNSS priors + local constraints + weighted graph optimization.
Project →Research pillar 03
Autonomous Systems & Navigation
How mobile and robotic systems perceive, localize and act when global positioning is unreliable and the environment is only partially known.
Scientific question
How should an autonomous machine build a usable model of an uncertain environment?
CoHMI combines global priors, local geometric constraints, remote sensing, sensor fusion and machine-aware environment models. Forest environments are a demanding test case because GNSS accuracy degrades, obstacles are irregular, and geometric free space does not automatically mean traversability.
Navigation stack
From map prior to vehicle action
Global prior→Localization→Local perception→Traversability→Dynamic cost map→Path planning→Vehicle control
Connected projects
Mapping, remote sensing and digital twins
Forest AI & Remote Sensing
Individual-tree detection, species classification and environment perception.
Project →Digital Twin of the Machine
Kinematic layer for manipulator motion coordination and machine-state representation.
Project →Digital Twin of the Working Field
Route, trees, obstacles, deadwood, terrain and local traversability.
Programme →Mobile Platforms
Operator-controlled and autonomous platform research across navigation and control.
Sensor fusion
Integration of GNSS, local geometry, remote sensing and onboard perception.
Selected publications
Published foundations
Hybrid Forest Mapping for Robotic Navigation Under GNSS Uncertainty
ICR 2026 · Springer LNCS 16790.
DOI ↗Review of Remote Sensing Data Processing Pipelines for Individual Tree Detection and Species Classification
Forests 17(10), 1134.
DOI ↗A Model-Based Kinematic Layer for Digital-Twin-Assisted Motion Coordination
Journal of Applied Engineering Science 24(3).
DOI ↗
Liudmila A. Steshina
Igor V. Petukhov
Evgenii N. Bogdanov
Ildar Grigorev
Daniil V. Galkin