There are a many sensors that can be used for localization
of an outdoor mobile agricultural robot. Common ones are
wheel encoders, GPS, inertial measurement unit (IMU) and
machine vision [1],[2]. Usually these are used in combination
to provide a global position estimate. This position can be
fairly accurate if expensive sensors are used.
On an agricultural robot, the global position does not
always have to be that accurate when operating on a field.
More important are the robot’s position relative the crops, if
the aim is agricultural operation on crop-scale. This requires
an ability to navigate locally at high precision [3]. An
example is a robot having a camera sensor detecting and
classifying crops, and an active tool performing individual
plant operations out of sight of the camera,