SocialDriveGen: Generating Diverse Traffic Scenarios with Controllable Social Interactions

2025-12-01

AI Summary

Existing traffic scene generation methods suffer from insufficient realism and diversity, particularly neglecting the influence of social preferences on driving behavior. To address this, we propose a hierarchical generative framework that jointly performs semantic reasoning and social preference modeling, representing driver personality and interaction style along an “egoistic–altruistic” dual-dimensional continuum to enable socially grounded, controllable scene generation. Our method integrates social-dimension representation learning with context-aware trajectory sequence modeling, trained and evaluated on Argoverse 2. Experiments demonstrate high-fidelity, diverse scene generation spanning cooperative to adversarial interactions, significantly enhancing the robustness and generalization of autonomous driving policies under rare and high-risk scenarios. The core contribution is the first explicit, interpretable modeling of social preferences as latent control variables in traffic generation—enabling personalized, adjustable simulation of interactive driving behaviors.

Technology Category

• Knowledge Representation and Reasoning: Preferences
• Natural Language Processing: Generation
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

The generation of realistic and diverse traffic scenarios in simulation is essential for developing and evaluating autonomous driving systems. However, most simulation frameworks rely on rule-based or simplified models for scene generation, which lack the fidelity and diversity needed to represent real-world driving. While recent advances in generative modeling produce more realistic and context-aware traffic interactions, they often overlook how social preferences influence driving behavior. SocialDriveGen addresses this gap through a hierarchical framework that integrates semantic reasoning and social preference modeling with generative trajectory synthesis. By modeling egoism and altruism as complementary social dimensions, our framework enables controllable diversity in driver personalities and interaction styles. Experiments on the Argoverse 2 dataset show that SocialDriveGen generates diverse, high-fidelity traffic scenarios spanning cooperative to adversarial behaviors, significantly enhancing policy robustness and generalization to rare or high-risk situations.

Problem

• Generates diverse traffic scenarios with controllable social interactions
• Models egoism and altruism to simulate varied driver personalities
• Enhances autonomous driving policy robustness in rare situations

Innovation

• Hierarchical framework integrates semantic reasoning and social preference modeling
• Models egoism and altruism as complementary social dimensions for controllable diversity
• Generates diverse high-fidelity traffic scenarios from cooperative to adversarial behaviors