Autonomous Robotic Crop Technology & Intervention Systems
Abstract
The goal of this project is the research, implementation and evaluation of autonomous technologies pertaining to a modular robotic system that can fully autonomously perform mechanical hoeing. The first innovative concept is the application of an image synthesis framework for generating large data set for training deep learning classifiers for agricultural purposes, allowing for novel methods of crop localization, characterization and re-identification of individual plants. The second innovative concept is the real-time augmentation of SLAM generated maps with semantic segmentation output, allowing for the development of enhanced obstacle detection. The final innovative concept is enhanced path- and mission-strategies that allow for raising and lowering of a robotic hoeing implement, crop row following and turning the vehicle at the end of crop rows with the final aim of achieving full field coverage. Finally, the developed methods are integrated on a robotic platform, evaluated and demonstrated in fully autonomous field tests.
Project staff
Reinhard Neugschwandtner
Assoc. Prof. Mag. DI Dr.nat.techn. Reinhard Neugschwandtner
reinhard.neugschwandtner@boku.ac.at
Tel: +43 1 47654-95117
Project Leader
01.02.2019 - 31.12.2020
BOKU partners
External partners
AIT Austrian Institute of Technology GmbH
Dr. Gerardus Croonen
coordinator