Adversarial and Reactive Traffic Entities for Behavior-Realistic Driving Simulation: A Review

TL;DR

The paper tackles the validation gap in autonomous vehicle planning caused by simulations that fail to capture realistic, reactive, and adversarial traffic. It introduces a taxonomy of dynamic entity behaviors (Replay, Rule-based, Learning-based, Language-based) and scenario control levels (No, Partial, Full) to structure evaluation in simulation. It surveys microscopic simulators and traffic datasets, and reviews a broad spectrum of reactive and adversarial traffic methods, including IL/RL, diffusion-based generation, and language-guided approaches. The work highlights key challenges—realism, diversity, controllability, and closed-loop testing—and outlines open questions and directions to advance robust AV planning validation in realistic traffic conditions.

Abstract

Despite advancements in perception and planning for autonomous vehicles (AVs), validating their performance remains a significant challenge. The deployment of planning algorithms in real-world environments is often ineffective due to discrepancies between simulations and real traffic conditions. Evaluating AVs planning algorithms in simulation typically involves replaying driving logs from recorded real-world traffic. However, entities replayed from offline data are not reactive, lack the ability to respond to arbitrary AV behavior, and cannot behave in an adversarial manner to test certain properties of the driving policy. Therefore, simulation with realistic and potentially adversarial entities represents a critical task for AV planning software validation. In this work, we aim to review current research efforts in the field of traffic simulation, focusing on the application of advanced techniques for modeling realistic and adversarial behaviors of traffic entities. The objective of this work is to categorize existing approaches based on the proposed classes of traffic entity behavior and scenario behavior control. Moreover, we collect traffic datasets and examine existing traffic simulations with respect to their employed default traffic entities. Finally, we identify challenges and open questions that hold potential for future research.

Figures (1)

• Figure 1: Proposed classification. In the absence of control over the driving scene, all entities replay their logged trajectories, while the AV tries to maintain a safe state. In a partial controllability setting, individual entities are selected for control by a behavior function, while all other entities are set to replay. Finally, under full controllability, all entities except the AV are controlled by a single or multiple behavior functions.