Abstract:
The problem of modeling the routes of self-driving vehicles in a traffic flow in which there are no collisions is being solved. A new swarm algorithm based on a microscopic model of traffic flow is proposed, which ensures the movement of agents without collisions. Changes in several optimality criteria during the operation of the algorithm are considered, such as: average speed of agents, throughput, number of lane changes. The boundaries of the effective values of the hyperparameters of the algorithm are estimated. At certain density parameters and push/pull coefficients in the traffic flow, free flow and an improvement in the values of the optimization criteria are observed.
Keywords:swarm intelligence, swarm algorithm, microscopic model of traffic flow, agent, fish school algorithm.