A new study demonstrates containerised AI deployment on mobile agricultural robots using the AgriGaia platform, offering reproducible weed detection, modular workflows and remote management, advancing precision farming and collaborative research across institutions.

The paper titled “Containerized AI Deployment on Mobile Robots Using the AgriGaia Platform” has been accepted for presentation at GIL 2026. Conducted by Moritz Gerwin, Volker Saggau and Floris Ernst at Universität zu Lübeck Institut für Robotik und Kognitive Systeme, the work describes how containerised training and Docker based deployment enable a Clearpath Jackal robot to identify Rumex weeds in pasture grass. Two YOLO models, one general detector and one specialised for Rumex, run on the robot with Portainer managing the containers. The AgriGaia data space provides open, collaborative tools for data ingestion, annotation, training and monitoring, ensuring reproducibility and data sovereignty in line with Gaia X principles.

The system leverages GPU acceleration, TensorRT optimisation and ONNX model exchange to achieve real‑time inference on edge devices. Field tests show a mean average precision of 0.86, highlighting the potential for scalable AI solutions in agriculture. The project “Robotik auf der Digitalen Weide” is funded by the European Agricultural Fund for Rural Development within the European Innovation Partnership EIP Agri Schleswig‑Holstein, underscoring the relevance of public support for digital farming innovations.

More info at https://gil-net.de/konferenzen/tagung-2026/