A Highway-Driving System Design Viewpoint Using an Agent-Based Modeling of an Affordance-Based Finite State Automata

Abstract

This paper presents an agent-based modeling framework for affordance-based driving behaviors during the exit maneuver of driver agents in human-integrated transportation problems. We start our discussion from one novel modeling framework based on the concept of affordance called the affordancebased finite state automata (AFSA) model, which incorporates the human perception of resource availability and action capability. Then, the agent-based simulation illustrates the validity of the AFSA framework for the highway-lane-driver system. Next, the comparative study between real driving data and agent-based simulation outputs is provided using the transition diagram. Finally, we perform a statistical analysis and a correlation study to analyze affordance-based driving behavior of driver agents. The simulation results show that the AFSA model well represents the perception-based human actions and drivers' characteristics, which are essential for the design viewpoint of control framework of human driver modeling. This paper is also expected to benefit a designed control for autonomous/self-driving car in the future.

Keywords

• Agent-based modeling
• affordance
• finite state automata
• driving behavior
• human-machine interactions