A main issue in the design of a stabilizing feedback controller for robotic mining excavators is the compensation of unknown forces which are exerted on the excavator's dipper when removing the soil. These forces may vary according to the condition of the soil while harsh operating conditions make difficult their direct measuring with the use of dedicated force sensors. To address this problem the article proposes a flatness-based disturbance observer and control method for the dynamic model of thee 2-DOF robotic mining excavator. It is proven that the dynamic model of the robotic excavator is differentially flat and thus it can be transformed into the input-output linearized form and into the canonical Brunovsky form. The forces which are generated due to the contact of the robot's dipper with the soil take the form of additive disturbance inputs. Next, the state-vector of the robot is extended by considering as additional state variables the cumulative disturbance terms and their time-derivatives. The new model is observable, thus a Kalman Filter-based disturbance observer is finally designed which allows for estimating simultaneously the state variables of the robot and the additive perturbation inputs. By identifying precisely the forces which are due to the contact with the ground one can also update the system's flatness-based controller so as to ensure global stability and fast convergence of the state variables of the robotic excavator to the associated setpoints.