BioTraffic: a bio-inspired behavioral model to vehicle traffic simulation

• Main Authors: de Quadros, Carlos Eduardo Pereira, Adamatti, Diana Francisca, de Lima Bicho, Alessandro
• Format: Conference Proceeding
• Language: English
• Subjects: agent-based systems; Analytical models; behavioral models; Biological system modeling; Computational modeling; Games; lane changing; lane merging; Merging; Space vehicles; terrain reasoning; traffic simulation; Visualization
• ISSN: 2159-6662
• DOI: 10.1109/SBGames54170.2021.00014

This paper presents a new microscopic model to simulate the behavior of vehicle traffic through a bio-inspired agent-based method. The proposed model reinterprets a biologically-motivated method for generating leaf venation patterns in order to propose terrain reasoning, a technique that has been widely used in simulations and games. The main idea is to represent unoccupied spaces through abstract markers distributed in the environment. These markers identify free regions for the movement of vehicles in the simulated traffic environment. The markers provide space information and report the vehicular flow of the simulated scenario, including flow density and velocity. Typical behaviors observed in real traffic, including inhomogeneous driver models, lane changing and merging trajectories, are emergent properties of the proposed model. We demonstrate the flexibility and robustness of our model on simulation environments, comparing the statistical results with a commercial software used for traffic simulation.