In this report, a differential-driven rover
for use in agriculture is analyzed with the
end goal of determining and controlling its
position.
For determining the rover’s position as well
as its other states, an extended Kalman
filter has been designed combining multiple
sensors to achieve higher precision than
what would have been achievable by using
only one sensor. The sensors used are wheel
encoders, an inertial measurement unit, a
magnetometer, and a GNSS with RTK
capabilities.
The system states are used for designing
two different kinds of controllers. The first
one is a heading controller, to ensure that
the rover keeps the correct heading while
driving along at a constant linear velocity.
The second is an inertial frame controller,
ensuring that the rover reaches predefined
positions.
The extended Kalman filter as well as all
the sensor input are handled by a microcontroller running FreeRTOS to ensure realtime operation. While the control and interface to the motors are handled by an
NVIDIA Jetson Nano running Robot Operating System.