Derivative-free distributed nonlinear Kalman filtering for cooperating agricultural robots

The paper proposes state estimation-based control for cooperating agricultural robots. To estimate with accuracy the position of the robotic vehicles as well as their motion characteristics, fusion of estimates from multiple sensors is performed with the use of the Derivative-free distributed Kalman Filter. The proposed derivative-free nonlinear filtering method enables distributed state estimation, by substituting the Extended Information Filter with the standard Information Filter recursion. This filtering approach has significant advantages because, unlike the Extended Information Filter, it is not based on local linearization of the nonlinear dynamics and computation of Jacobian matrices. The proposed nonlinear control is in accordance with the principles of differential flatness theory. The performance of the considered distributed filtering-based control is tested through simulation experiments on the problem of autonomous navigation of agricultural robots under a master-slave scheme.