[Paper Review] Human-Like Decision Making for Autonomous Driving: A Noncooperative Game Theoretic Approach

Peng Hang, Chen Lv+3 more|arXiv (Cornell University)|May 22, 2020

Traffic control and management40 references18 citations

TL;DR

This paper proposes a human-like decision-making framework for autonomous vehicles (AVs) using noncooperative game theory—specifically Nash equilibrium and Stackelberg games—to model interactions with human-driven vehicles. By integrating potential field methods and model predictive control for motion prediction and planning, the framework enhances safety, comfort, and efficiency, with the Stackelberg approach reducing decision-making cost by over 20% compared to Nash equilibrium in lane change scenarios.

ABSTRACT

Considering that human-driven vehicles and autonomous vehicles (AVs) will coexist on roads in the future for a long time, how to merge AVs into human drivers traffic ecology and minimize the effect of AVs and their misfit with human drivers, are issues worthy of consideration. Moreover, different passengers have different needs for AVs, thus, how to provide personalized choices for different passengers is another issue for AVs. Therefore, a human-like decision making framework is designed for AVs in this paper. Different driving styles and social interaction characteristics are formulated for AVs regarding driving safety, ride comfort and travel efficiency, which are considered in the modeling process of decision making. Then, Nash equilibrium and Stackelberg game theory are applied to the noncooperative decision making. In addition, potential field method and model predictive control (MPC) are combined to deal with the motion prediction and planning for AVs, which provides predicted motion information for the decision-making module. Finally, two typical testing scenarios of lane change, i.e., merging and overtaking, are carried out to evaluate the feasibility and effectiveness of the proposed decision-making framework considering different human-like behaviors. Testing results indicate that both the two game theoretic approaches can provide reasonable human-like decision making for AVs. Compared with the Nash equilibrium approach, under the normal driving style, the cost value of decision making using the Stackelberg game theoretic approach is reduced by over 20%.

Motivation & Objective

• To enable autonomous vehicles to integrate smoothly into human-driven traffic by emulating human driving behaviors.
• To address the challenge of AVs' misfit with human drivers, minimizing disruptions in mixed-traffic environments.
• To provide personalized decision-making that accounts for diverse passenger preferences regarding safety, comfort, and efficiency.
• To develop a decision-making framework that models dynamic interactions between AVs and human-driven vehicles using game theory.
• To evaluate the framework’s effectiveness in realistic, high-stakes driving scenarios such as merging and overtaking.

Proposed method

• Formulates driving styles and social interaction characteristics into decision-making models, balancing safety, comfort, and efficiency.
• Applies Nash equilibrium and Stackelberg game theory to model noncooperative interactions between AVs and human-driven vehicles.
• Integrates potential field methods for motion prediction to estimate the behavior of surrounding traffic agents.
• Employs model predictive control (MPC) for real-time motion planning, using predicted trajectories as inputs.
• Combines game-theoretic decision-making with MPC and potential fields to generate human-like, adaptive responses in dynamic traffic.
• Uses predicted motion data from the potential field and MPC modules to inform strategic decisions in game-theoretic frameworks.

Experimental results

Research questions

• RQ1 How can autonomous vehicles make decisions that closely mimic human driving behavior in mixed-traffic environments?
• RQ2 What game-theoretic approach—Nash equilibrium or Stackelberg—better balances safety, comfort, and efficiency in AV decision-making?
• RQ3 To what extent can the integration of motion prediction and planning improve the realism and effectiveness of AV decision-making?
• RQ4 How do different driving styles and passenger preferences influence the design and performance of AV decision-making frameworks?
• RQ5 Can the proposed framework achieve measurable improvements in decision-making cost and human-likeness in real-world driving scenarios?

Key findings

• The Stackelberg game-theoretic approach reduced the decision-making cost by over 20% compared to the Nash equilibrium approach under normal driving conditions.
• Both Nash equilibrium and Stackelberg game models produced human-like decisions, demonstrating feasibility in complex traffic interactions.
• The integration of potential field methods and model predictive control enabled accurate motion prediction and real-time planning, supporting effective game-theoretic decisions.
• The framework successfully handled two critical lane change scenarios—merging and overtaking—showcasing robustness in high-stakes driving contexts.
• The decision-making framework effectively balanced safety, ride comfort, and travel efficiency, aligning with human driver preferences.
• The results confirm that game-theoretic modeling significantly improves AV behavior in mixed-traffic environments, reducing unnatural or disruptive maneuvers.