Claus G. Smitt

Tel: +49 – 228 – 73 – 2597
Office: Nussallee 5, EG, room 0.014
University of Bonn
Agricultural Robotics & Engineering, ILT
Nussallee 5
53115 Bonn

Short CV:

Claus Smitt is a Research Assistant and Ph.D. Candidate at the University of Bonn, working in the Agricultural Robotics & Engineering department of the Institut für Landtechnik since January 2020. He received his Electrical Engineering Degree from Universidad Nacional de Rosario in 2014 and his Master’s degree from Instituto Balseiro in 2016, both in Argentina. For the next, 2 years he worked developing robotic inspection systems for the nuclear industry at CNEA also in Argentina. During 2019 he worked for iRobot in Pasadena, CA on developing Visual SLAM & Sensor fusion algorithms for consumer robots.

Research Interests:

  • Agriculture Robotics
  • 3D Crop Mapping
  • Machine Learning
  • Autonomous Navigation


  • Python Applied to Machine Learning, SS 2020


Multi-modal Crop Detection and Quality Assessment

The increasing need of automating agricultural tasks presents new opportunities and challenges to the field of robotics. In particular, the use of image sensors for crop detection and mapping opens the possibility of producing yield, quality, and ripeness estimations, among others, which are of great use for farmers.
Convolution neural networks (CNN) had a surge in popularity in the last decade, becoming the de-facto approach for image classification, detection & segmentation. In order to further improve their accuracy in challenging scenarios, such as fruit/crop detection in agricultural environments, several imaging modalities can be combined.
This project evaluates several CNN architectures that fuse multiple information sources (RGB, Stereo, NIR, LiDAR, etc.) to produce better crop detection that can be used for estimating the aforementioned metrics, and as input for 4D crop mapping and intervention systems.

Automated Horticulture Crop Surveying Platform

With the aim of better understanding crop growth and development in glasshouses in general and specific plantations over time, the Agricultural Robotics group is developing an automated Crop Surveying platform. It consists of a retrofitted OTS glasshouse crop management platform, equipped with an extensive sensor suite and a robot arm on top.
One of the noteworthy sensors is an RGB-D+NIR camera array which is later combined to achieve a FOV as large as the surveyed crops. This enables scanning a whole plantation row on a single run, producing high-quality depth-aware datasets. This data is key for crop reconstruction and semantic mapping, as well as yield, quality, and ripeness estimation.
Moreover, the robot arm can potentially be employed for other surveying tasks that require close proximity or interaction with the crop or even perform crop intervention.
This system is currently surveying experimental sweet pepper and tomato plantations at the University of Bonn’s Campus, Klein Altendorf.