Agriculture Reference
In-Depth Information
constantly changing as it travels to fulfill farming tasks, and a good estimation of
heading in real time is essential for map coherence. With the new (real-time) position
of the estimates comprising the position of the vehicle in
E
-
N
-
D
coordinates and its
heading angle, any vehicle navigating on flat terrain can compose a
global map
from
multiple
local maps
generated by the awareness system of the vehicle. The geometry
involved in this transformation is schematized in Figure 12.6 for a point
P
, whose
local coordinates deduced, for example, with a stereo camera are (
x
p
,
y
p
) when the
camera is globally referenced by position (
e
c
,
n
c
), and the traveling direction of the
vehicle forms an angle
with the east axis. When terrains are not flat and the three
attitude angles need to be taken into account, the transformation is more complex
(Rovira-Más 2011b), but the philosophy is exactly the same.
The procedure described above to build global maps from perceptual data can be
followed, both
offline
and
on the go
while the vehicle is traveling, as long as instanta-
neous position and attitude are readily available and precise enough. Unfortunately,
this will not always be the case as sensors may fail or deliver noisy outputs unexpect-
edly. In such a situation, the global map construction must cease adding perceptual
data, and consequently blank areas with no information will appear in the map;
nevertheless, void areas can always be filled in subsequent missions as long as data
being added is consistent. Let us take the important case of GPS receivers. Precision
and consistency greatly differs from low-cost mass-produced receivers to profes-
sional units set to implement differential corrections. Differential receivers (DGPS)
represent an effective way to compensate for atmospheric errors, but multipath
reflections and signal blockage induced by vegetation are sometimes inevitable. At
times, the
number of satellites
detected by the receiver or the
dilution of precision
ϕ
N
n
p
P
φ
O
LOCAL
(
e
c
,
n
c
)
e
c
n
p
n
c
O
GLOBAL
(0, 0)
e
p
E
e
p
FIGURE 12.6
Construction of global maps from perceptual data.
