Fahrspureinstiegs- und Ausstiegsvorhersage für Autonome Arbeitsfahrzeugnavigation

Details

Veröffentlichungs-Nr.

EP4748216

Anmeldetag

11. November 2025

Veröffentlichung

27. Mai 2026

Rechtsraum

EP

Abstract

Systems and techniques are disclosed for predicting lane entry and exit points to guide an autonomous work vehicle operating in fields of tree rows or similar environments. Range sensors capture depth and intensity data, which a trained segmentation model classifies into trunks, canopy, ground, and obstacles. Trunk centroids are clustered and used to estimate lane geometry and a lane centerline. Canopy points are analyzed in down-lane bins to compute density measures, enabling detection of row-end locations. Lane exit points are projected from canopy features onto the lane centerline, and a primary exit point is selected to support headland turns. Control modules generate steering and speed commands that guide the vehicle toward the exit point while monitoring position, speed, and trajectory. The process repeats continuously at sensor update rates, allowing real-time detection of row ends and smooth autonomous transitions between lanes.

Anmelder

Firma

Deere & Company

Land

🇺🇸 USA