Donate to Science & Enterprise

S&E on Mastodon

S&E on LinkedIn

S&E on Flipboard

Please share Science & Enterprise

Drones, A.I. Deployed by Agricultural Start-Up

Philipp Lottes and Cyrill Stachniss

Philipp Lottes, left, and Cyrill Stachniss with one of their drones. (Barbara Frommann, University of Bonn)

24 Sept. 2019. A new company is combining drone photography with artificial intelligence to provide growers with better data to monitor and manage crop growth. The company, Pheno-Inspect, based In Bonn, Germany, began in June 2019, spun-off from the Photogrammetry and Robotics Lab at University of Bonn.

The Photogrammetry and Robotics Lab studies robotics in agriculture, including the use of autonomous robotics and big-data analytics to help growers make critical decisions brought on by climate change. The lab’s PhenoRob project uses land-based sensor networks with ground and aerial drones to collect data. The project is also developing analytical models for evaluating crop phenotypes or characteristics, with the goal of promoting more sustainable practices and maximizing crop output in times of more demanding climatic conditions.

Pheno-Inspect is the creation of doctoral candidate Philipp Lottes, commercializing his research with robotics professor and lab director Cyrill Stachniss. The company is developing a technology using aerial drones hovering 10 to 100 meters off the ground with high-resolution cameras to take photos of crops in the field, transmitted to a database for analysis. That analysis employs machine learning to evaluate the phenotypes, or characteristics, of crops as they appear in the images. With that analysis, farmers can precisely assess different varieties or effects of factors, such as fertilizer or weather, on crops.

“Global population growth means that agriculture will have to produce even higher yields in the future, while the area of the arable land remains the same,” says Lottes in a university statement. “The current bottleneck in the development of new and better varieties is high-throughput phenotyping in the field.”

Lottes, Stachniss, and lab colleagues describe the technology in a paper published last month in the Journal of Field Robotics (paid subscription required). The paper outlines a system using drones to capture detailed images of planted fields, with the ability to discriminate between crops and weeds. The images are then fed into a database trained with machine learning to identify and characterize detailed elements of crops, for example stems and leaves.

The machine-learning algorithm is trained with a convolutional neural network that combines image analysis and machine learning to dissect an image by layers for understanding features in the image. Different aspects of each layer discovered and analyzed by the algorithm are translated into data that the algorithm then uses to train its understanding of the problem being solved, with that understanding enhanced and refined as more images and data are encountered.

The analysis, say the researchers, can provide detailed crop data for individual crop rows or specific field locations. The team tested the system on different fields and crops in different countries, and conclude that the technology can be deployed to completely new fields unseen before by the system.

While Pheno-Inspect only began in June 2019, the company already attracted financial support of €270.000 ($US 298,000) from the Start-Up University Spin-Offs program of the state of North Rhine-Westphalia in Germany and the European Union.

More from Science & Enterprise:

*     *     *

Comments are closed.