Abstract
Actual and upcoming climate changes will evidently have the largest impact on agriculture crop cultivation in terms of reduced harvest, increased costs, and necessary deviations from traditional farming. The aggravating factor for the successful applications of precision and predictive agriculture is the lack of granulated historical data due to slow, year-round cycles of crops, as a prerequisite for further analysis and modeling. A methodology of plant growth observation with the rapid performance of experiments is presented in this paper. The proposed system enables the collection of data with respect to various climate conditions, which are artificially created and permuted in the encapsulated design, suitable for further correlation with plant development identifiers. The design is equipped with a large number of sensors and connected to the central database in a computer cloud, which enables the interconnection and coordination of multiple geographically distributed devices and related experiments in a remote, autonomous, and real-time manner. Over 40 sensors and up to 24 yearly harvests per device enable the yearly collection of approximately 750,000 correlated database entries, which it is possible to independently stack with higher numbers of devices. Such accumulated data is exploited to develop mathematical models of wheat in different growth stages by applying the concepts of artificial intelligence and utilizing them for the prediction of crop development and harvest.
Highlights
Artificial intelligence (AI) today has a significant focus on increasing the efficiency of different sectors and reducing negative impacts on the environment
There are hundreds of phenological stages of wheat growth (BBCH scale) [18], and the system separates the data for three generalized ones: (i) germination, (ii) plant formation and maturation, and (iii) grain maturation, which are general approximated stages for many plants, with the open possibility for the concept to be transferable to other species
Accurate data about isolated and artificially created climatic conditions and consequent plant development are archived and, over time, are used to build a large data set of 6 million records collected about 5000 climate scenarios over a period of two years, which is suitable for applying algorithms in mathematical modeling and predicting future plant developments upon the onset of climate change
Summary
Artificial intelligence (AI) today has a significant focus on increasing the efficiency of different sectors and reducing negative impacts on the environment. The issue of the slow collection of large amounts of relevant historical data is tackled here by creating specially designed bioreactors that serve as rapid plant-model identification systems for multiple simultaneous climate zones, supported by autonomous real-time data acquisition and archiving. The observed outputs are: stem segment length, yield, grain number and size, and spectra observed in an incremental way with regards to historical trends, aimed to be captured by high-resolution and multi-spectral cameras, analyzed, archived, statistically processed, and, modeled by ML algorithms. This is the main premise of the project AgroSPARC–Smart and Predictive.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
