Abstract: The paper considers the problems of tactical situations recognition in the context of autonomous use of military robotic systems. In the opinion of most Russian and foreign experts, arming of troops with autonomous military robotic systems, equipped with reconnaissance assets and weapons, can significantly increase the effectiveness of conduct of operations and reduce depletion of ranks. The absence of an effective mechanism of tactical situations recognition in the context of the environment that is hard to formalize is one of the key deterrents to the creation of autonomous military systems. The authors suggest applying logical and linguistic methods and their neural network realizations, particularly, the multilayer perceptron, to solve the mentioned task. The modeling results demonstrate that the use of neural network realization of logical and linguistic methods of recognition, based on the multilayer perceptron, allows using such methods in the on-board computer intelligence of a military robotic system for the recognition of tactical situations. The use of the multilayer perceptron helps significantly speed up the process of the new knowledge processing, since in order to create the new system of tactical situations recognition it is necessary to merely correct the learning sample and retrain the multilayer perceptron; it takes less time than the formation of a new logical samples group by a group of experts.