EU: Automated monitoring of horticulture by spectral analysis with quantum dot detectors and high-resolution optical - HORTIQD

HortiQD aims to develop an affordable machine vision system for precision farming, specialized in horticulture. We will build a hyperspectral short-wave infrared (SWIR) camera working in the wavelength range of 1 to 2 µm. The sensory system will be used directly on the field, as in-vivo measurement, as either a handheld solution or attached to an autonomously driving tractor to allow for intensive monitoring of the crop.

The spectral analysis executed partially by artificial intelligence (AI) allows for the estimation of abiotic (of non-biological origin like water stress) and biotic (diseases and pests) stress to help farmers in the apples growing. The collected spectral data shall be correlated with additional information from sensors, usually already available at orchards, such as temperature and humidity. This monitoring is an element of Integrated Pest Management, which plays an important role to reduce the usage of chemical pesticides what is in line with “Farm 2 Fork” (F2F) European strategy and can constitute the element of a Decision Support System. The system will be a basis for a deduction of suitable measures and further investigations on how far the limits of spectral on-field-examinations can be pushed, e.g., regarding sugar and starch content of the fruits.

This processed data-set is the basis for increasing the yield, improving the quality of the harvest and further reducing the use of chemicals in the future. Cost-effective and close monitoring is the basis for an early warning system for diseases. Curbing them in their early stages contributes significantly to European sustainability goals in agriculture, by avoiding pesticides and protecting our food production systems. It is the explicit aim of the project to provide farmers with easy access to hyperspectral sensor systems and take advantage of the possibilities emerging with new technologies.

To reach this goal we will develop a novel detector type, based on quantum dot (QD) technology, promising much lower cost than state of the art detectors. This device will be combined with a customized tunable Fabry-Pérot interferometer to specifically select wavelengths and record spectra. The functional scope of the imaging apparatus will be extended by point analysis at specific fixed wavelengths in the visible and near infrared to fully cover the optimal spectral ranges necessary for a profound analysis of plant states and disease detection.

The design will consist of a large aperture for optimal light input for the hyperspectral imaging approach and full usage of the high spatial resolution of the QD detector. The system will be mounted to an autonomous tractor, in order to allow fully automated monitoring. The system-integration is strongly emphasized, in order to bring the solution close to the markets, including mounts, interfaces, hardware and software to process the spectral data and providing a professional infrastructure for data handling and correlation.

As a first case study the fruit with highest European impact will be addressed: the apple.

The following tables summarize the objectives and expected results we are aiming for, as well as suitable validation methods and the coherence with the call scopes. First the overall objective is presented and followed by more specific goals