The paper proposes a novel approach of transferring a free-flowing pile of different objects and materials significantly larger than the capabilities of the robot. Whilst trying to automatize feed-pushing in cattle farms with mobile robots, it was found, there may be no algorithm for mobile robots to do the job of this kind. With a robot measuring the impact force of his to the environment, it is possible to approximate how the overall mass pile is transferred as well as the overall change of mass. Change of the mass available to cattle is data that have not been available to farmers. This approach proposes that the feed pusher could give such information as a step closer to fully automatized cattle farms. As for feed pushers of nowadays, they are designed to drive the path hard-coded or even built into the farm, whereas this approach may give additional automation possibilities for the feed pusher, by using power consumption and position as a measurement, the proposed algorithm provides robot capabilities to approximate trajectories for the next passes and thereby be more effective. Approximations can be done by applying divide and conquer strategy to the pile itself, decomposing it in mass points with their relative area of action and the pre-design path it can move about. With this strategy it is proposed that the point is fixed to its path, but the mass is a variable, calculated based on the impact of a robot to one or multiple points of mass. The authors are proposing iterative algorithm to approximate the feed pushing robot trajectory based on sensed force feedback. The change of mass over time can be used to estimate the overall change in the pile's mass - actual consumption, as well as overall transfer over time. It is possible to describe most of the use-cases in the feed-pushing process, but there might be a wide range of other applications that include different free-flowing elements stacked in a pile that require reordering, gathering or transfer.