Skip to main content Skip to main navigation

Publication

A Tool Set for Development and Testing of Perception Systems in Agriculture

Jan Christoph Krause; Sebastian Röttgermann; Alexander Tauber; Jens Herbers; Naeem Iqbal; Mark Niemeyer; Hannes Hollmeier; Thorben Boße; Simon Schirk; Robin Hilker; Stefan Stiene; Martin Atzmueller
In: VDI Wissensforum GmbH (Hrsg.). VDI-Berichte. International Conference Agricultural Engineering (LAND.TECHNIK-2025), 82th International Conference on Agricultural Engineering LAND.TECHNIK AgEng 2025, November 7-8, Hannover, Germany, VDI Verlag, Düsseldorf, 2025.

Abstract

Reliable perception of the surroundings of an agricultural machine is a core requirement to ensure safe and efficient operation of automated agricultural machinery. The perception performance depends on factors like weather, vegetation, and operational conditions. When developing perception systems, performance must be evaluated for all operational conditions. Since this involves combinations of several environmental parameters, the design, management, and evaluation of relevant test scenarios is complex. Within the research project AI-TEST-FIELD, we build a comprehensive tool set to record comparable sensor data of different weather and vegetation conditions and connect that with an evaluation pipeline; with that, we can enhance that data with auto-generated ground truth and predictions from perception algorithms. Our toolset enabled us to record a broad data set with comparable test scenarios under different environmental conditions, allowing us to investigate the influence of weather and plants on the performance of sensor systems.