Evaluating Scenario-based Decision-making for Interactive Autonomous Driving Using Rational Criteria: A Survey

2025-01-03

AI Summary

This study addresses the lack of standardized evaluation criteria for cross-scenario performance assessment of deep reinforcement learning (DRL) algorithms in autonomous driving decision-making. We propose the DDTUI five-dimensional rationality evaluation framework—encompassing Safety, Efficiency, Training Efficiency, Altruism, and Explainability—and systematically benchmark mainstream DRL methods across four canonical interactive scenarios: highway driving, ramp merging, roundabouts, and unsignalized intersections. By establishing a scenario-classification modeling approach and a multi-criteria evaluation benchmark, we achieve the first quantitative, comparable, and engineering-oriented comprehensive assessment of DRL algorithms’ multi-objective performance. Results identify distinct strengths and bottlenecks of each algorithm across dimensions, and pinpoint three key optimization directions: enhancing model explainability, modeling altruistic behaviors, and improving generalization via robust training strategies. This work provides both theoretical foundations and practical guidelines for the trustworthy deployment of DRL in autonomous driving systems.

Technology Category

• Reasoning under Uncertainty: Sequential Decision Making
• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Security and Privacy: Large-scale security measurements

Abstract

Autonomous vehicles (AVs) can significantly promote the advances in road transport mobility in terms of safety, reliability, and decarbonization. However, ensuring safety and efficiency in interactive during within dynamic and diverse environments is still a primary barrier to large-scale AV adoption. In recent years, deep reinforcement learning (DRL) has emerged as an advanced AI-based approach, enabling AVs to learn decision-making strategies adaptively from data and interactions. DRL strategies are better suited than traditional rule-based methods for handling complex, dynamic, and unpredictable driving environments due to their adaptivity. However, varying driving scenarios present distinct challenges, such as avoiding obstacles on highways and reaching specific exits at intersections, requiring different scenario-specific decision-making algorithms. Many DRL algorithms have been proposed in interactive decision-making. However, a rationale review of these DRL algorithms across various scenarios is lacking. Therefore, a comprehensive evaluation is essential to assess these algorithms from multiple perspectives, including those of vehicle users and vehicle manufacturers. This survey reviews the application of DRL algorithms in autonomous driving across typical scenarios, summarizing road features and recent advancements. The scenarios include highways, on-ramp merging, roundabouts, and unsignalized intersections. Furthermore, DRL-based algorithms are evaluated based on five rationale criteria: driving safety, driving efficiency, training efficiency, unselfishness, and interpretability (DDTUI). Each criterion of DDTUI is specifically analyzed in relation to the reviewed algorithms. Finally, the challenges for future DRL-based decision-making algorithms are summarized.

Problem

• Deep Reinforcement Learning
• Autonomous Vehicles
• Decision Performance

Innovation

• Deep Reinforcement Learning
• Autonomous Driving
• Performance Evaluation