The article is devoted to the analysis of modern neuroprocessor architectures aimed at solving problems related to automatic control systems of mobile systems. The system approach for solving the problems of neuroprocessor architectures formation using the features and functions implemented in the form of a hardware part is considered. a comparative analysis of approaches to the implementation of neuroprocessor architectures in relation to systems of automatic control of mobile systems is evaluated. This article presents the architecture of neuroprocessors and neural network computers in relation to solving problems for strapdown inertial navigation systems (bins). This kind of tasks for modern on-board digital computers (bcvm) with traditional architecture is quite difficult to implement in them.