© vandenborneaardappelen
Author profile picture

Hands-free growing, autonomous cultivation, growing as a service, data-drive growing; if you’ve been following greenhouse horticulture for a while, you will come across these terms more and more often. Experts believe that this is the way to grow enough, healthy and safe food all over the world. I wrote about this earlier in The road to hands-free production of greenhouse crops‘.

But what exactly is hands-free or autonomous cultivation? Everyone in the sector has their own ideas about it. It may not seem important in itself, but time and again it turns out that it is useful to ‘speak the same language’. If a robot developer says to a market grower: ‘We will be able to supply the technology for autonomous cultivation in two years’ time’, both parties may have a different concept about that.


A grower may think that such a robot is capable of doing all the work independently, while the robot developer envisages a system that can do part of the work itself, but that an employee must always be present to carry out some of the tasks. Before you even realize it, both parties are dissatisfied. The grower does not want to buy the system after two years because it is not yet really fully autonomous, and the developer will have to keep on developing it for much longer before they can really start selling it.

An important part of the solution is found in ‘agile’ development through the robot builder and the grower working together. Determine where the value lies in the development process, develop, test and adjust the development based on what you have learned together. This is exactly what we see happening in various projects revolving around the AgriTech vouchers. A camera apparently is unable to capture things precisely enough, the system is not practical when you have to work with it in the greenhouse for a whole day, or the greatest value of a certain product is just at a slightly different place in the growing process.

Common language

However, anyway, even though you are working together this way, having a common language is still very important. You are dealing with several stakeholders, after all. As a grower you are not only dealing with technology suppliers, the government and your employees, but also with the insurance company, for instance. In talks with all these different parties it is important to have the same thing in mind when talking about autonomous cultivation.

In order to make a start with this common language, Priva and TU Delft AgTech Institute have asked a team of students to develop a framework for the classification of autonomous cultivation in greenhouse horticulture. It is explicitly not the intention of both parties to set one-sided standards, they want to help the entire sector to map out and further encourage the developments around autonomous cultivation.

Levels of autonomy

As a basis for the framework, we have been looking at the framework surrounding autonomous driving. The most recent version of the framework for autonomous driving is SAE J3016_202104. Similar to this framework for autonomous driving on the road, 6 levels of autonomy have also been defined within the framework for autonomous cultivation: 

0 – No automation: the grower and its crew is fully responsible for all activities.

1 – Grower Assistance: The system assists the human.

2 – Partial Automation: Grower supervises (“Hands Off”).

3 – Conditional Automation: Grower is fallback (“Eyes Off”).

4 – High Automation: Grower sets goals (“Brains Off”).

5 – Full Automation: All operational phases at Level 4 and integrated together.

Where autonomous driving can distinguish between driving on the highway, an enclosed area or parking, the framework for autonomous cultivation has also distinguished between different (main) tasks. To do this, the work has been divided into four dimensions:

1 – Logistics: Moving products; sorting according to quality and size and packaging.

2 – Crop Manipulation: Physical actions at crop level; harvesting; pruning and trellising.

3 – Crop Growing: Assessing plant state; creating climate setpoints and controlling the climate inside the greenhouse.

4 – Greenhouse Strategy: High level management decisions; scheduling phases.

These six levels and four dimensions can be used to describe the degree of autonomy for a greenhouse horticultural company. As an example there is the trial with autonomous crops of grower Rijk Zwaan and Blue-Radix. This involves the ‘climate control’ component. 

Autonomous greenhouse

In the trial, the grower adjusted the ‘setpoints’ once a week at the ‘autonomous’ greenhouse. The Blue-Radix Crop Controller then took care of ensuring that the required values were met. Normally this is a process in which the grower themselves checks and adjusts the values several times a day. This process requires a grower who has the right kind of knowledge about the crop, their greenhouse and climate parameters.

Now, based on historical data, measurements from the greenhouse and the weather forecast, the algorithm saw to it that the correct values were achieved. In this case, we are talking about level 4 automation, humans act as a back-up and under normal circumstances will not need to worry about the regulation of the climate.


By establishing the framework, an important step has been taken towards the further development of autonomous growing in covered crops. Having a better idea of how autonomy can be classified also allows for a more informed conversation about what still needs to be developed technologically-wise. Furthermore, it can contribute to research on the impact beyond the technical developments. Consider, for example, who is to be held accountable if things go wrong? What tasks remain for humans in an autonomous greenhouse. More on that in a future column.

As explained, Priva and TU Delft AgTech Institute want to use the newly developed framework to initiate the development and improvement of an industry-wide applied ‘language’ for autonomous greenhouse horticulture. The use of the framework as a standard and its continued development is currently being discussed with a number of industry parties. The framework can be downloaded here

Photo: Autonomous agricultural vehicle

About this column:

In a weekly column, alternately written by Eveline van Zeeland, Eugene Franken, Katleen Gabriels, PG Kroeger, Carina Weijma, Bernd Maier-Leppla, Willemijn Brouwer, Maarten van Andel and Colinda de Beer, Innovation Origins tries to figure out what the future will look like. These columnists, sometimes joined by guest bloggers, are all working in their own way to find solutions to the problems of our time. So tomorrow will be good. Here are all the previous articles.