Farming robots collecting plant breeding data
Being used more and more on production farms, farming robots collecting plant breeding data start paving their way to the research department.
When monitoring the plants growth, researchers are gathering information to characterize every experimental condition. Yield and harvest quality are useful results but most of the time the study is more complex, it focuses on subtler trait expression. The information required can be collected manually by observing the plants and taking notations, or by farming robots collecting plant breeding data !
Farming robots can “learn” complex processes and achieve precision tasks in production fields. They replace manual labor and lessen agrochemical use, therefore reducing costs and environmental impact. Such robots also scan the production fields, gathering images to map field heterogeneities, and can as well be used to detect anomalies and apply treatment directly.
In the plant breeding department and in research stations, can be found other kinds of farming robots collecting plant breeding data in the fields. Integrating this new data set to the existing research information system can be complex and challenging. How can the IT system be adapted to integrate, store and analyze all this data?
1. Farming robots in the production fields
In production farms, farming robots can replace hard work such as fruit picking, pruning or chemical sprayings: by weeds destruction or drone auxiliaries dropping.
• Grass management robots for vineyards and forestry such as Vitirover
• Soil preparation with the high precision geopositioning CEOL by Agreenculture
Will these herds of mowers replace herbicides in the fields? It may soon become a reality, and we’ll see more and more help from robots, starting with fruit crops since they require manual work to maintain the trees and pick fruits, such as pruning bots.
Research departments worldwide are currently studying such farming robots, either to create new ones or to study the benefits of existing models on specific crops.
2. Robots collecting plant breeding data
The BoniRob is an agricultural robot developed by Bosch that can conduct autonomously repeating phenotyping tasks for crop stands and even for individual plants. Furthermore, it can be used as a carrier, supplier and base for multiple BoniRob-Apps. Current apps are (i) phenotyping, (ii) penetrometer and (iii) precision spraying.
The vineyard mapping Physiocap by Fruition Sciences that measures shoot diameter, number of shoots per vine and average biomass, in each row it scans and generates maps.
The Anatis robot by Carré is an example of auto-driving soil preparation robot able to collect several measures on the soil, air and plants.
“Sensor-embedded farming robot collecting plant breeding data, which are especially useful for plant breeders, who have to painstakingly analyze thousands of plants for plant size and color, fruit size and form, and insect damage” Bosch Deepfield Robotics experts.
3. High throughput phenotyping robots
This area of research consists in gathering as many pictures and measures of the studied plants as possible, generating a big data set that can increase the efficiency of crop improvement. Phenotyping platforms (HTPPs) such as Phenopsis enable such study indoor, but farming robots collecting breeding data are providing similar services to field breeding and testing:
The autonomous Phénomobile platform that collects plant information for evaluating growth, nutrition and stress data.
Heliaphen by Sunrise is a high throughput phenotyping robot to study plant responses to draught stress from germination to maturity.
The autonomous Robotanist by Carnegie Mellon University’s Robotics Institute provides high resolution spatial, spectral, and physical information about Sorghum field data.
4. Robots VS other monitoring tools
Various methods enable plant researchers to studying trait expression during the plant growth. Let’s compare them according to two criteria, for comparable resources :
• DEBIT: How much information they provide over time.
• EXHAUSTIVITY: How many plots can be covered over a period of time.
The graph on the left provided by Agri Sud-Ouest Innovation shows that farming robots collecting plant breeding data can be more efficient than manual data collection, and drones even more.
Nevertheless, every method has specific advantages, like the accuracy of still weather stations, the analytical ability of technicians taking notations, the homogeneity of robot measures, and the capacity of drones to capture data instantly over a large area. It’s a combination of such factors that influence the choice of researchers for the data collection method that fits optimally its objectives.
“when you follow the protocol with the robot, you hardly have to go back”
Jérôme Pain, Deputy Director of Operations at Bejo
5. Plant breeding robots in the R&D IT system
Arvalis researchers in phenotyping reckon the biggest challenge is computational: “The change in obtained data type raises the question of their management and their integration in breeding tools and decision-making tools”.
When it comes to implementing new data and processes at the heart of the researcher’s work tool, precautions must be taken and every project adapted to its existing system. Besides, it is crucial to be able to adapt the R&D IT system to integrate the wealth of field information provided by automatic phenotype observation.
An interface needs to be built between the farming robot collecting plant breeding data and the existing experimentation management database. On the left is a typical example of data processing cycle Doriane has studied for one of its customers using aerial drones (drone data in green).
Implementation methods and tools may vary a lot, depending on the research objectives and the type of robot data or work needed, and may vary as well in time, since these technologies are evolving fast. Therefore, only research departments with an evolutive IT system will be able to integrate plant breeding robots.
To go further..
By the way, here is the link to watch our Webinar about Integrating drones to the R&D system
Integrating drones to the R&D system
Innovative phenotype data collected by drones enable agronomists to study new traits and take better decisions:
→ Disease follow-up trait: Measure the plots' disease sensibility indicator
→ Yield-related traits: Leaf area index, height, QPAR (total amount of solar energy intercepted)
→ Counting plants, leaves, fruits