Аннотация:The paper describes the development, testing, and implementation of navigation algorithms for unmanned wheeled agricultural vehicles when the navigation equipment consists of a low-precision inertial navigation system (INS) with microelectromechanical sensors (MEMS), a receiver of Global Navigation Satellite Systems (GNSS), and odometry data — odometer-derived velocity and a steering sensor.In the operation of unmanned agricultural machinery, the navigation problem is one of the most important. For the treatment of most agricultural crops, the positioning accuracy of the corresponding equipment needs to be a few centimetres. High accuracy is also important to reduce fuel costs and ensure safe operation of agricultural machinery for humans. The reliability of the navigation solution depends on its stability to outliers and short-term losses of GNSS data.The work considers two types of equipment configuration which are structurally different: non-articulated (classic) and articulated. The classic configuration implies a four-wheeled vehicle, which is steered by the front or the rear axle. The articulated configuration implies two frames connected by a joint, each of which has two wheels fixed to the corresponding frame. In this configuration, one can make a turn by changing the angle between the frames.