Le 03/06/2020

Patent mapping analysis in the field of agricultural robotics

This paper presents the first preliminary results of a mapping study of patents in agricultural robotics. A more complete presentation has been provided during the International Forum of Agricultural Robotics on December 11, 2019. However, these first results already come up with a view on the sector and demonstrate the exploration capabilities of patent mapping tools through three examples of angle of attack (by filing date, by country, by company).

All invention patents are classified and identified by industrial property offices. The classification methodology is an international standard defined by the World Intellectual Property Organization. This international patent classification offers a unique reading of technological fields and makes it possible to rank inventions of the same nature according to common rules for all countries in the world.

Patents are all published 18 months after filing, which results in an extensive amount of accessible technical information, some of it (active patents) entitling monopoly rights to their holders. Patent databases contain up to 140 million systematically classified technical documents. This structured and public data can be analyzed using big data tools.


The “patent landscaping” or patent mapping is a global interactive landscape that uses a topographic map-like representation (with reliefs) to illustrate the proximity between the different technical concepts. The high quality of the patent database used by the INPI combined with the power of a big data processing algorithm (Naive Bayes classification), allows the reliable identification of patents dealing with the same subject. This makes it easier for a non-specialist public to access the patent competitive space rather than other modes of representation. Regarding agricultural robotics, we thus obtain a map similar to a geographic map on which we can see areas of patent concentration illustrated by mountains whose height and width symbolize the density of filings in the field.

The research strategy used to produce these maps requires a combination of International Patent Classification entries, as well as keywords specific to the agricultural robotics sector. The analysis covers the last 20 years, which is the maximum term of protection offered by a patent.

Fig.1. Mapping of the field of agricultural robotics

At the end of this strategy, 11,000 patent families - all the patents filed by the same applicant in different countries to protect the invention - were identified, thanks to the software that distributes them on the map. A large part of these families has been filed recently (half between 2015 and 2019). The strategy chosen for the pre-study highlighted 2,248 patents published in 2018, compared to only 989 in 2016. The sector's filing dynamic has therefore increased significantly over the past three years.

The mapping reveals 10 areas of patent density which correspond to areas where the research and development activity, and therefore the patent filing, has been somewhat more intense. This includes the White Mountain areas identifiable on the map by their white level line. The following major areas can therefore be distinguished in decreasing order of volume:

1.Robotic arms (for handling, harvesting, etc.)

2.Means of travel and specialized vehicles

3.Mechatronics applied to fishing or fish farming

4.Animal husbandry and analysis

5.Image capture

6.Cutting tools

7.RFID tags applied to agriculture

8.Plant growing systems

9.Robotic supply of food

10.Automated irrigation


The map shows areas of lower concentration corresponding to most emerging areas. Recent filings are located in these areas of lower patent density. Examples include irrigation techniques, mowing equipment, techniques for optimizing soil fertilization, electronics dedicated to the agricultural field (sensor, communication). This corresponds on the map to the hilly areas (medium-sized mountains) where the patents were filed in 2018 or 2019.

Fig.2. Highlighting patents filed in 2018 and 2019


The two largest patent applicants in the field of agricultural robotics are the United States (in green) and China (in red). They account for half of the patent applications in the agricultural robotics sector over the past 20 years. But interestingly, these applications focus on specialized areas with, on the one hand, many American filings in agricultural machinery (vehicle and tools) and on the other hand, a strong Chinese presence in the field of robotic arms, and preferably the electronics applied to the sector (sensors, communication module, RFID tags ...). Moreover, the patent applications of these two countries are polarized on different fields with little overlap.

Fig.3 Highlighting Chinese (red) and American (green) patents


The first French filings are far less well represented with only a hundred patent families distributed in several places on the map. Portfolios are more concentrated on cutting tools and RFID. Some isolated patents in the blue areas representing the sea and on the edge of hills reflect technologies that depart from the usual issues and are therefore potentially original and innovative.

Fig.4. Highlighting patents filed in France


John Deere is considered as a key player in agricultural robotics and a leading company in the field. Mapping made it possible, for example, to take a closer look at the patenting strategy of this company.

Fig.5. Highlighting patents filed by John Deere

The highlighting of the company's patents reveals a portfolio on the upper left side of the map that is concentrated in the fields of agricultural vehicles and harvesting tools. In addition, the company is diversifying beyond agricultural machinery with patents on robotic devices for sowing, harvesting or spraying, on cropping systems and agricultural data processing methods.

Thanks to the INPI mapping service, all these patents can be identified more precisely, extracted on request, and categorized in more detail. Other highlighting can be carried out to provide a strategic reading of the evolution of the field, the competition or the academic research. This information generally helps complete the strategic vision of the leaders to enable them to make decisions that identify the risks and opportunities in conducting an entrepreneurial activity.

These first results are delivered as raw data and may require processing, in particular to refine the keywords presented on each mountain. Noise filtering may also be required and will help to determine more meaningfully the trends identified in this first analysis

Project presented in the scientific seminar organised by Robagri during the 2019 FIRA:


Arnaud Gourragne
Anas Hanaf
Erwan Chapelier

Mapping Tool

Clarivate Analytics, from Derwent Innovation.


Patent base:  https://bases-brevets.inpi.fr/fr/accueil.html  
Mapping of INPI inventions:  https: // www .inpi.fr / en / cartographie-des-inventions  
International Patent Classification:  https://www.wipo.int/classifications/ipc/fr 

Categories : #Markets