|
|
|
| Authors: | M. Kawollek, T. Rath |
| Keywords: | robotic harvesting, image processing, Gerbera jamesonii |
Abstract:
A system for automatic harvesting of Gerbera jamesonii cut flowers grown in greenhouses was developed using image processing.
For the harvesting process of flower stems an industrial robot with six axes was used mounted on an additional linear axis.
For harvesting an end-effector was developed using razor blades for cutting the stems.
Special calibration algorithms were used for online calibration of the whole harvesting system (robots, end effectors, transport unit, cameras). For the harvesting process, results of image processing from eight differ-ent viewpoints where analysed to enable the identification of all flower stems in at least one of the positions.
Harvesting was performed stepwise, if necessary in differ-ent positions of the plant.
To realise a collision-free harvesting process a path-planning module was integrated.
Additionally, an algorithm for “flower stem track-ing” was implemented to predict the emerging of the flower stem in the following none-visible area.
In harvesting experiments 80 % of all flower stems where har-vested.
The rate of harvested flower stems was decreasing with an increasing num-ber of flower stems per plant.
Looking at plants with one or two flower stems 98 % of the flower stems where harvested. 51 % of all flower stems where harvested if plants have five ore more flower stems.
|
Download Adobe Acrobat Reader (free software to read PDF files) |
|