Abstract in English:
While the mechanical weeding in the inter-row is current
practice, the removal within the row is a challenging issue.
Automatic systems have been studied based on crop detection
or weed recognition, but these techniques are expensive and
with variable reliability.
During the first year, the poplar growth in the short rotation coppice
(SRC) is rather slow and, hence, in order to avoid remarkable
yield losses, the control of intra-row weeds assumes a
fundamental importance. Within the Italian project Suscace,
a simplified detection system, consisting of a photoelectric (PS)
and a capacitive sensor (CS) has been designed and tested in
order to assess the degree of accuracy of the poplar cuttings
detection along the row.
A one-year old poplar stand sited in Central Italy was used
for the study. The distance between the rows was 3 m with
a spacing of 0.50 m between the cuttings within the row. The testing apparatus consisted of a small platform moving
along a monorail four meter long. This was placed parallel to
the planted row and its detection capacity appraised at three
distances (0.15, 0.20, and 0.25 m) from the row and two
speeds (1.0 and 1.5 km h-1), values routinely adopted during
mechanical weeding.
Compared to the PS, the CS revealed a higher amplitude
of the response in presence of a poplar cutting. The CS,
once identified the plant, allowed its precise confinement, an
information helping to integrate the slight interference (due the
non-uniformity in shape, size and development of the cuttings)
observed for the PS. The signals of PS and CS complement
each other by bridging the respective gaps in the identification
of the cuttings. Interestingly, any significant differences were
observed for different distances and speeds. |