Аннотация:When designing an autonomous vehicle, the size, weight, and power consumption, as well as cost (SWaP-C) of an onboard computer are strictly limited. At the same time, modern algorithms of localization, computer vision, and mapping, which must be run to ensure autonomy, can have high computational complexity. This article proposes a method, which analyzes SWaP-C factor as well as a performance of modern onboard computers to establish restrictions on the space of computers that meet the required limits. To achieve that, the dataset of existing embedded computers is collected, and the mathematical model of onboard computers is suggested. Based on the proposed model, the Pareto optimal set is found. It is shown that the size of the set of computers required to be considered after optimization is on average five times smaller than the initial set.