Evidence and quantification of cooperation of driving agents in mixed traffic flow

Abstract

Cooperation is a ubiquitous phenomenon in many natural, social, and engineered systems with multiple agents. Understanding the formation of cooperation in mixed traffic is of theoretical interest in its own right, and could also benefit the design and operations of future automated and mixed-autonomy transportation systems. However, how cooperativeness of driving agents can be defined and identified from empirical data seems ambiguous and this hinders further empirical characterizations of the phenomenon and revealing its behavior mechanisms. Towards mitigating this gap, in this paper, we propose a unified conceptual framework to identify collective cooperativeness of driving agents. This framework expands the concept of collective rationality from our recent model (Li et al. 2022a), making it empirically identifiable and behaviorally interpretable in realistic (microscopic and dynamic) settings. This framework integrates mixed traffic observations at both microscopic and macroscopic scales to estimate critical behavioral parameters that describe the collective cooperativeness of driving agents. Applying this framework to NGSIM I-80 trajectory data, we empirically confirm the existence of collective cooperation and quantify the condition and likelihood of its emergence. This study provides the first empirical understanding of collective cooperativeness in human-driven mixed traffic and points to new possibilities to manage mixed autonomy traffic systems.