Capturing the swarm intelligence in truckers: The foundation analysis for future swarm robotics in road freight

Abstract

Abstract A group of individual truckers can be regarded as a swarm intelligence system without central management. With the development of autonomous driving technology, trucker groups will be replaced by driverless vehicles. At that point, a swarm of truckers will become a swarm robotics system. Therefore, considering the design and control of an efficient swarm robotics system, it is essential to investigate the properties and model the behaviors of a swarm of truckers in advance. In this study, we probe the characteristics of both individual truckers and a swarm of truckers using trajectory data of truckers. First, the trajectory data were map matched based on the geographic scale of cities and administrative regions. Then, the properties of the division of labor, pattern formation, and swarm synchronization were obtained through an analysis of the spatiotemporal distribution of radius of gyration, travel distance, and the number of visited places. Because predicting the next visit locations of individuals of a swarm is a measure for modeling swarm behaviors, the prediction model can be used to predict future swarm robotics (driverless trucks) behaviors. Thus, we apply several machine learning models to predict the next locations of truckers. The results show that there are common characteristics and routines embodied in the behavior of the truckers; the swarm shows consistency and regularity. Moreover, the peak predictability of the entire group reached 94%, indicating that our model can predict the behavior of groups and individuals. Our findings provide basis supporting to the future efficient swarm robotics system.