Where markers show what autonomous driving in the greenhouse requires
In greenhouse horticulture, an autonomous vehicle often drives along fixed aisles. As soon as a vehicle also has to carry out other tasks between crop rows, light, crop, viewing angle and space immediately matter. In Use Case 12, the project examined whether Aruco markers, a type of QR code for navigation, can help Octiva’s autonomous vehicle with such routes. The test shows when markers can be detected well and where they fall short for accurate positioning. This allows you to read the result as knowledge about greenhouse conditions, not as a ready-made solution.
Innovation package, use case and type test
Greenhouse Horticulture
Robot Platform
Status: completed
Technical functionality
Broad knowledge question
What is needed to move beyond fixed aisles
As an entrepreneur, cultivation manager or operator, you mainly know autonomous driving from fixed aisles. Between crop rows, fixed infrastructure elements cannot simply be installed. New implements require different routes through the greenhouse than moving row by row through the crop. The project therefore examined whether flexible elements, such as visual markers, can support autonomous navigation throughout the greenhouse. For you, the value lies in seeing whether such an aid fits the space, crop and row layout in which the vehicle has to operate.
Approach
Markers tested under controlled conditions and in the greenhouse
The researchers used Aruco markers, a type of QR code for navigation. They tested different conditions, including lighting, viewing angle, marker position in relation to the greenhouse infrastructure and the influence of the crop. Using a known reference, they determined whether the markers were clearly visible and how accurately the vehicle could determine its position. They also compared controlled conditions with the real greenhouse at Tomatoworld. This shows which conditions matter before markers can be assessed as an aid for localisation.
Objective
Seeing where marker recognition differs from precise positioning
The test had two objectives. The researchers wanted to assess how well Aruco markers can be recognised under operational conditions. They also determined the positioning accuracy that can be achieved with these markers. For you, that distinction matters because a visible marker does not automatically mean that the vehicle knows exactly where it is.
Result and reflection
Markers help with localisation but do not solve positioning on their own
According to the test, Aruco markers can help with localisation. At the same time, they are not sufficient on their own for good positioning accuracy. Within the Handsfree Agrifood ecosystem, this is a useful outcome because autonomous navigation in the greenhouse requires knowledge about the conditions in which a vehicle can see, drive and orient itself. You see where visual markers can support navigation and which limits first need to be clear before use between crop rows can be assessed further.
Successful
You can see from the test which combination of size, number of cells and texture worked best for the marker.
You see that a camera with a wide light range is important, while resolution is less critical.
You see that standard lenses were sufficient in this test.
You see that depth measurements were generally fairly accurate.
Lessons learned
You need the camera to view the marker plane perpendicularly for accurate XYZ positioning.
You see that localisation works better when the marker is placed higher rather than close to the ground.
You see that image sharpness must be good, but was not critical in this test.
You can assess Aruco markers as an addition for localisation, not as a sufficient basis for good positioning accuracy.