Specific features of methods and algorithms for planning unmanned vehicles' routes in dynamically changing road scene

Abstract

In this work the authors tackle the question of unmanned navigation and calculate the route in the traffic flow. With the growth of vehicle autonomy this issue becomes systemic and requires upgrading of the existing models, methods and algorithms since some of the decisions normally made by a person (driver) are now transferred to the on-board unit of movement modeling and decision-making unit. This work presents methods for calculating the foute of an unmanned vehicle route based on the information about various levels of the transport system. The 1st level describes the generalized characteristics of the global route based on the chosen set of criteria; 2nd level describes the position and movement dynamics for the multitude of the traffic participants in the vicinity; level 3 shows the calculated trajectory for the vehicle maneuvering in the traffic flow. The suggested methods were developed with the use of multi-agent multi-level simulation model of the transport system proposed by the authors earlier, which allows for route calculation based on a variety of criteria.