Agriculture Reference
In-Depth Information
(b)
(a)
FIGURE 7.39
(a) Autonomous vehicle in operation within citrus grove, ladar, and vision
field of view. Vision systems tracking path and canopy in-row. (b) End of row detection.
(From Subramanian, V. et al.,
Trans. ASABE
, 52, 5, 1-12, 2005;
Comput. Electron. Agric.
,
53, 130-143, 2006. With permission.)
They instrumented a John Deere e-gator as shown in Figure 7.38 and implemented a
Fuzzy Enhanced Kalman Filter to fuse ladar, vision, and IMU path estimates to predict
optimal course correction efforts. Through the use of a Pure Pursuit algorithm, steering
and speed commands were generated to drive the position error to zero. This approach
provided more robust path correction than previous unfused sensory data selected on a
priority basis in combination with a PID-based steering controller. In addition to in-row
navigation, a vision- and ladar-based approach was developed for accommodating vari-
ous uncertainties such as missing trees and end-of-row conditions. Figure 7.39 shows the
e-gator moving down the row with vision and ladar constantly scanning the forward
looking terrain. By observing abrupt changes in ladar range to tree canopy and variation
in path versus tree spectral characteristics, algorithms were developed that could detect
the end of row conditions. A visual odometry-based dead reckoning approach was imple-
mented to assist end-of-row turning, thus providing the capacity for fully autonomous
navigation within the grove.
7.6 NOVEL TECHNOLOGIES FOR CROP STATUS MONITORING
During the past decade, numerous studies have sought to develop various forms of
crop status monitoring using technologies such as machine vision in the visible and
near-infrared spectral regions, laser radar, ultrasonics, and more recently, biosen-
sors. These sensor technologies are being used to monitor crop factors such as yield
estimation, canopy size, canopy volume, leaf density, disease pressures, and pest
pressures. These data are then being used in various forms of precision agriculture
to control fertilization cost, chemical usage, and even irrigation demands. In this sec-
tion, we will briefly introduce several of the applications under development.
7.6.1 L
ASER
U
SED
IN
P
RECISION
S
PRAYER
Campoy et al. (2010) established a tree canopy modeling and precision spraying
system by using laser sensors and 3-D maps. Chen et al. (2011) developed a LIDAR
