A novel framework for modelling traffic flows with the participation of unmanned ground vehicles

Abstract

The article presents a new software platform for modelling traffic flows involving unmanned vehicles, using a number of advanced technological solutions, in particular, the FLAME GPU supercomputer agent modelling framework, intelligent software modules based on fuzzy and hierarchical clustering, genetic optimization algorithms, a subsystem for visualizing the state of agents-vehicles based on OpenGL, etc. As a result, it is possible to simulate the dynamics of road accidents and output traffic for various configurations of digital road networks within the “smart city” scale. The possibility of using the proposed software platform (implemented using C++ and FLAME GPU) is demonstrated for the problem of estimating the sensitivity of the total number of accidents in relation to important control parameters: arrival rate and average speed of unmanned ground vehicles interacting with usual ground vehicles and other agents of the intelligent transport system.