Artificial Intelligence for Modeling and Simulation of Mixed Automated and Human Traffic

Saeed Rahmani; Shiva Rasouli; Daphne Cornelisse; Eugene Vinitsky; Bart van Arem; Simeon C. Calvert

cs.AIApr 14, 2026

Artificial Intelligence for Modeling and Simulation of Mixed Automated and Human Traffic

Authors: Saeed Rahmani, Shiva Rasouli, Daphne Cornelisse, Eugene Vinitsky, Bart van Arem, Simeon C. Calvert

Organizations: Department of Transport & Planning, Delft University of Technology, Delft, the Netherlands. · Department of Industrial and Systems Engineering, University of Michigan, Dearborn, Michigan, USA · 3Tandon School of Engineering, New York University, New York, United States.

Abstract

Autonomous vehicles (AVs) are now operating on public roads, which makes their testing and validation more critical than ever. Simulation offers a safe and controlled environment for evaluating AV performance in varied conditions. However, existing simulation tools mainly focus on graphical realism and rely on simple rule-based models and therefore fail to accurately represent the complexity of driving behaviors and interactions. Artificial intelligence (AI) has shown strong potential to address these limitations; however, despite the rapid progress across AI methodologies, a comprehensive survey of their application to mixed autonomy traffic simulation remains lacking. Existing surveys either focus on simulation tools without examining the AI methods behind them, or cover ego-centric decision-making without addressing the broader challenge of modeling surrounding traffic. Moreover, they do not offer a unified taxonomy of AI methods covering individual behavior modeling to full scene simulation. To address these gaps, this survey provides a structured review and synthesis of AI methods for modeling AV and human driving behavior in mixed autonomy traffic simulation. We introduce a taxonomy that organizes methods into three families: agent-level behavior models, environment-level simulation methods, and cognitive and physics-informed methods. The survey analyzes how existing simulation platforms fall short of the needs of mixed autonomy research and outlines directions to narrow this gap. It also provides a chronological overview of AI methods and reviews evaluation protocols and metrics, simulation tools, and datasets. By covering both traffic engineering and computer science perspectives, we aim to bridge the gap between these two communities.

Figures & tables

Fig. 1: Survey methodology, scope, and structure

Fig. 2: Taxonomy of AI methods for modeling mixed-autonomy traffic

Survey	Mixed Autonomy Focus	Traffic Agent Model.	Multi-Agent Methods	World Models	Scenario Generation	Cognitive / Physics	Sim. Tools & Datasets	Cross- Discipl.
Schwarting et al. [ 22 ]	–	(✓)	–	–	–	–	–	–
Kuutti et al. [ 27 ]	–	–	–	–	–	–	✓	–
Grigorescu et al. [ 28 ]	–	(✓)	–	–	–	–	(✓)	–
Chao et al. [ 31 ]	(✓)	✓	(✓)	–	–	–	✓	(✓)
Di and Shi [ 21 ]	✓	–	(✓)	(✓)	–	(✓)	–	✓
Kiran et al. [ 29 ]	–	–	–	–	–	–	(✓)	–

TABLE I: Comparison of this survey with related surveys across key dimensions. ✓ = covered; (✓) = partially covered; – = not covered.

Method	Training Signal	Loop	Strengths	Limitations	Representative Studies
Imitation Learning / Behavior Cloning	Expert demonstrations (supervised)	OL, CL	Simple and scalable; direct data-to-policy mapping; strong baselines with large datasets	Covariate shift and compounding errors; causal confusion; limited robustness to out-of-distribution states	ALVINN [ 14 ] , PilotNet [ 15 ] , CIL [ 16 ] , ChauffeurNet [ 34 ] , DAgger [ 35 ] , TransFuser [ 36 ] , Urban Driver [ 37 ]
Trajectory Prediction	Logged trajectories (supervised)	OL	Multi-modal forecasting; mature standardized benchmarks; captures distributional structure of future motion	Marginal prediction; non-reactive; displacement metrics miss interactive quality	CoverNet [ 38 ] , MultiPath [ 39 ] , TNT [ 40 ] , DenseTNT [ 41 ] , LaPred [ 42 ] , PRIME [ 43 ] , MID [ 12 ]
Reinforcement Learning	Reward signal (trial-and-error)	CL	Discovers novel strategies via exploration; no expert demonstrations needed; inherently reactive	Sample inefficiency; reward engineering difficulty; safety during exploration; sim-to-real gap	Kendall et al. [ 18 ] , Roach [ 44 ] , BC-SAC [ 45 ] , Think2Drive [ 46 ] , HAIM-DRL [ 47 ] , Safe RL [ 17 ]
IRL / Adversarial Imitation	Expert demos (adversarial matching)	CL	Recovers interpretable reward functions; robust to compounding errors; captures driving preferences	Training instability; mode collapse; sensitivity to discriminator design; scaling difficulty	MaxEnt IRL [ 48 ] , GAIL [ 49 ] , Kuefler et al. [ 50 ] , PS-GAIL [ 51 ] , AIRL [ 52 ] , RAIL [ 53 ]
End-to-End Learned Planners	Expert demos + perception labels	OL, CL	Joint perception–planning; end-to-end gradient flow; unified architecture	Ego-status bias; open-loop metric gap; limited closed-loop generalization	NMP [ 54 ] , UniAD [ 55 ] , VAD [ 56 ] , SparseDrive [ 57 ] , InterFuser [ 58 ] , ThinkTwice [ 59 ] , NAVSIM [ 60 ]
Model-Based Policy Learning	Learned world model + reward	CL	Sample-efficient via imagination; decouples dynamics learning from policy optimization	Model bias; compounding errors in imagined rollouts; sensitive to model accuracy	World Models [ 61 ] , Dreamer [ 62 ] , DreamerV3 [ 63 ] , MILE [ 64 ] , AdaWM [ 65 ] , Imagine-2-Drive [ 66 ]

TABLE II: Summary of single-agent behavior modeling methods for mixed autonomy traffic simulation.

Method	Core Idea	Strengths	Limitations	Representative Studies
Joint Trajectory Forecasting (predominantly open-loop)
Joint Trajectory Prediction	Predicts joint future trajectories for multiple agents; uses graph networks, attention, or transformers to model inter-agent dependencies	Captures inter-agent dependencies; multi-modal outputs; strong benchmark performance; joint consistency	Does not model reactive feedback; joint consistency not guaranteed under marginal losses; gap to interactive simulation	Social LSTM [ 88 ] , Social GAN [ 89 ] , DESIRE [ 90 ] , Trajectron++ [ 91 ] , PRECOG [ 92 ] , VectorNet [ 93 ] , LaneGCN [ 94 ] , Scene Transformer [ 95 ] , HiVT [ 96 ] , MTR++ [ 97 ] , GameFormer [ 98 ]
Centralized Neural Simulation (closed-loop)
Diffusion-Based Simulators	Models multi-agent evolution via iterative denoising; samples from learned joint trajectory distributions with guidance for controllability	High diversity; multi-modal generation; controllable via guidance; collision-aware sampling	Slow sampling (iterative denoising); hard constraint satisfaction difficult; closed-loop stability over long horizons	MotionDiffuser [ 13 ] , CTG [ 99 ] , CTG++ [ 100 ]
Token-Based / Autoregressive	Discretizes motion into tokens; casts simulation as conditional language modeling with next-token prediction	Scalable transformer architectures; handles variable agent counts; real-time inference possible	Discretization error; compounding token mistakes; serialization order asymmetry; constraint satisfaction requires constrained decoding	MotionLM [ 101 ] , Trajeglish [ 102 ] , BehaviorGPT [ 103 ] , SMART [ 104 ]
Regression-Based Reactive	Single learned model produces globally consistent scene updates via continuous regression; rolls forward step-by-step	Global scene consistency; captures multi-agent dependencies; computationally efficient; responds to ego deviations	Long-horizon drift; limited multi-modality (point estimates); counterfactual validity concerns from passive logs	SimNet [ 105 ] , TrafficSim [ 106 ]

TABLE III: Summary of multi-agent interaction methods for mixed autonomy traffic simulation.

Method	Output Space	Core Idea	Strengths	Limitations	Representative Studies
Generative World Models
Video-Based & Latent World Models	Multi-view images; latent tokens; video frames	Learns action-conditioned video or latent generation via autoregressive token prediction or latent diffusion; produces future visual observations for end-to-end evaluation	Photorealistic rollouts; end-to-end stack evaluation; controllable via actions, text, and structured inputs; emergent scene dynamics	Long-horizon drift; high computational cost; causal consistency under intervention not guaranteed; image quality metrics miss driving-critical correctness	GAIA-1 [ 138 ] , GAIA-2 [ 19 ] , Drive-WM [ 139 ] , DriveDreamer [ 140 ] , DriveDreamer-2 [ 141 ] , Vista [ 142 ] , Copilot4D [ 143 ] , MUVO [ 144 ] , DrivingWorld [ 145 ] , GenAD [ 146 ]
Occupancy-Based World Models	3D/4D semantic occupancy grids; BEV fields	Predicts future semantic occupancy conditioned on ego actions via autoregressive or diffusion generation in voxel or triplane space	Geometric and semantic structure; directly supports planning cost evaluation; captures free-space and drivable-area evolution	No agent identity or intent; compounding topology errors; memory-intensive voxel grids; expensive 4D labels	OccWorld [ 147 ] , OccSora [ 148 ] , DOME [ 149 ] , T 3 Former [ 150 ] , $I^{2}$ -World [ 151 ] , Drive-OccWorld [ 152 ] , OccLLaMA [ 153 ]
Traffic Scenario Generation
Scene Initialization & Scenario Synthesis	Static scene layouts; multi-agent trajectories; map + traffic configurations	Generates traffic configurations ranging from static actor placements to full dynamic scenarios; controllable methods expose control interfaces (language, temporal logic, guidance functions, retrieval tags) for steerable generation	Data-driven scene diversity; user-specified controllability; intuitive language interfaces; compositional control; supports scenario augmentation and targeted testing	Realism–control trade-off; stronger constraints push off data manifold; unstandardized control interfaces; scene initialization methods lack dynamic validation; limited closed-loop validation	SceneGen [ 154 ] , TrafficGen [ 155 ] , DriveSceneGen [ 156 ] , CTG [ 99 ] , CTG++ [ 100 ] , Scenario Diffusion [ 157 ] , LCTGen [ 158 ] , ChatSim [ 159 ] , RealGen [ 160 ]
Safety-Critical Generation & Adversarial Falsification	Scenario parameters; adversarial agent trajectories; safety-critical multi-agent configurations	Discovers failure-inducing scenarios via optimization-based search (RL adversaries, importance sampling, fuzzing) or safety-targeted generative synthesis (guided diffusion, latent-space optimization with learned traffic priors)	Efficient failure discovery; targeted stress testing; identifies safety-critical edge cases; generative methods maintain realism via learned priors; complements naturalistic evaluation	Validity–criticality trade-off; adversarial drift to implausible scenarios; coverage gaps; failure finding $\neq$ risk estimation; guided generation provides soft not hard constraint satisfaction	AST [ 161 , 162 ] , Accelerated Eval. [ 163 ] , STRIVE [ 164 ] , KING [ 165 ] , AdvDiffuser [ 166 ] , CAT [ 167 ] , AV-FUZZER [ 168 ] , CCDiff [ 169 ] , CaDRE [ 170 ] , BridgeGen [ 171 ] , MADS [ 172 ]

TABLE IV: Summary of neural simulation environment and scene generation methods for mixed autonomy traffic simulation.

Method	Theory Basis	Core Idea	Strengths	Limitations	Representative Studies
Bounded Rationality and Decision-Making Models
Prospect-Theoretic Decision Models	Cumulative Prospect Theory (CPT); Prospect Balancing	Replaces expected utility with S-shaped value function and nonlinear probability weighting; captures loss aversion and overweighting of rare catastrophic outcomes in driving decisions	Interpretable style variation via CPT parameters; reproduces conservative/aggressive profiles without separate rewards; psychologically validated	Parameter calibration from driving data is difficult; limited to single-decision settings; does not model temporal dynamics of choices	Tversky & Kahneman [ 181 ] , Schmidt et al. [ 182 ] , Sun et al. [ 183 ]
Risk Allostasis & Homeostatic Models	Risk Allostasis Theory; Task-Capability Interface	Drivers adjust behavior to maintain a subjective target risk level rather than optimizing utility; explains systematic deviations from time-optimal behavior	High fidelity to naturalistic car-following and speed choice; captures risk compensation effects; simple parameterization	Target risk level is latent and hard to estimate; limited to longitudinal control; does not explain strategic interaction	Fuller [ 184 ] , Mohammadian et al. [ 185 ] , Kashifi [ 186 ]
Drift-Diffusion Models (DDMs)	Sequential Sampling; Evidence Accumulation	Decision variable accumulates noisy perceptual evidence toward boundaries; models the temporal dynamics and variability of gap acceptance, braking, and yielding decisions	Captures full RT distribution, including dangerous late decisions; perceptual inputs (looming, TTC) as natural drift rates; differentiable neural implementations	Binary decisions only; limited multi-alternative extensions; requires careful specification of evidence signals; calibration from naturalistic data is challenging	Ratcliff [ 187 ] , Markkula et al. [ 188 ] , Zgonnikov et al. [ 189 ] , McDonald et al. [ 190 ] , Fengler et al. [ 191 ]
Level- $k$ Reasoning	Cognitive Hierarchy; Behavioral Game Theory	Agents have heterogeneous strategic depth: level-0 follows heuristics, level- $k$ best-responds to level- $(k{-}1)$ ; models realistic variation in anticipatory reasoning	Principled heterogeneity in interactive behavior; avoids full-rationality assumption; calibratable from observed interactions	Requires specifying level-0 heuristic; distribution over levels is scenario-dependent; computational cost grows with $k$	Stahl & Wilson [ 117 ] , Chong et al. [ 118 ] , Li et al. [ 192 ]
Cognitive Architectures, Attention, Risk Perception, and Trust Models

TABLE V: Summary of cognitive and physics-informed AI methods for mixed autonomy traffic simulation.

Fig. 3: Timeline of model development in driving simulation and virtual evaluation

Dataset	Location(s)	Year	Duration / Scenarios	Freq. (Hz)	Environment	Agent Types	Key Features
Drone-Based Trajectory Datasets
NGSIM [ 227 ]	CA, GA (USA)	2005–06	90 min	10	Highway, arterial	Vehicles	Foundational; car-following analysis
highD [ 226 ]	Germany	2018	16.5 h / 110k vehicles	25	Highway (6 sites)	Cars, trucks	High precision ( $<$ 10 cm); 5,600 lane changes
inD [ 228 ]	Germany	2020	10 h / 11.5k road users	25	Urban intersections (4 sites)	Vehicles, cyclists, pedestrians	Naturalistic intersection behavior
rounD [ 229 ]	Germany	2020	6 h / 13k road users	25	Roundabouts (3 sites)	Vehicles, cyclists, pedestrians	Complex interaction patterns
pNEUMA [ 231 ]	Athens (Greece)	2018	12.5 h / 0.5M trajectories	25	Urban CBD (congested)	Cars, taxis, buses, PTWs, medium/heavy vehicles	Swarm of 10 drones over 5 days; dense multimodal trajectories over large urban network

TABLE VI: Major publicly available datasets for trajectory prediction, behavior modeling, and planning in automated driving. Duration indicates total recording time; scenarios indicates number of extracted segments for benchmarking where applicable.

Tool	Primary Application	Sim Level	Supported Datasets	Background Agents	Throughput
State-Level / Data-Driven Simulators
GPUDrive	Multi-agent RL training	State	WOMD	Learned policies, rule-based	Very high (GPU-native, 1M+ steps/s)
Waymax	Motion planning, prediction evaluation	State	WOMD	Log replay, IDM, learned	High (JAX/GPU/TPU batching)
nuPlan Devkit	Planning benchmark and evaluation	State	nuPlan	IDM, log replay, learned (SMART)	Moderate (multi-process)
Nocturne	Multi-agent RL research	State	WOMD-derived scenarios	Log replay, learned (BC baseline)	High (C++ core, $>$ 2k steps/s)
highway-env	RL algorithm prototyping	State	Synthetic (procedural)	IDM, MOBIL	High (vectorized environments)

TABLE VII: Simulation tools commonly used with learning-based driving methods. Simulation level indicates output granularity: sensor (camera/LiDAR images), state (bounding boxes, trajectories), or traffic (aggregate flow). Background agents refers to how non-ego vehicles are modeled during simulation.

Explore similar work

Aug 28, 2024

TrafficGamer: Reliable and Flexible Traffic Simulation for Safety-Critical Scenarios with Game-Theoretic Oracles

While modern Autonomous Vehicle (AV) systems can develop reliable driving policies under regular traffic conditions, they frequently struggle with safety-critical traffic scenarios. This difficulty primarily arises from the rarity of such scenarios in driving datasets and the complexities associated with predictive modeling of multiple vehicles. Effectively simulating safety-critical traffic situations is therefore a crucial challenge. In this paper, we introduce TrafficGamer, which facilitates game-theoretic traffic simulation by viewing common road driving as a multi-agent game. When we evaluate the empirical performance across various real-world datasets, TrafficGamer ensures both the fidelity, exploitability, and diversity of the simulated scenarios, guaranteeing that they not only statically align with real-world traffic distribution but also efficiently capture equilibria for representing safety-critical scenarios involving multiple agents compared with other methods. Additionally, the results demonstrate that TrafficGamer provides highly flexible simulations across various contexts. Specifically, we demonstrate that the generated scenarios can dynamically adapt to equilibria of varying tightness by configuring risk-sensitive constraints during optimization. We have provided a demo webpage at: https://anonymous.4open.science/api/repo/trafficgamer-demo-1EE0/file/index.html.

Guanren Qiao, Guorui Quan, Jiawei Yu +2

Chinese University of Hong Kong, Shenzhen, Guangdong, 518172 China · University of Manchester, Manchester, M13 9PL England · Shenyang MXNavi Co., Ltd, Shenyang, 110169 China

Autonomous Driving Simulation Autonomous Driving Scenario Generation

Ask PDF

Jun 12, 2026

ReactSim-Bench: Benchmarking Reactive Behavior World Model Simulation in Autonomous Driving

Reactive capability is a key property of data-driven behavior world model simulators for autonomous driving simulation systems. With this capability, simulated world agents can respond feasibly to autonomous vehicle (AV) behaviors that differ from the log. However, existing behavior simulation benchmarks do not directly measure reactive capability. They often let the simulator jointly control the AV and surrounding agents and evaluate realism through log similarity or open-loop prediction metrics. In this work, we introduce ReactSim-Bench for evaluating the reactive capability of behavior world model simulation in autonomous driving. We decouple the control of agents and the AV, using AV behaviors that differ from the log and require agents to respond as independent AV inputs. To obtain these AV behaviors, we construct a pipeline that uses an AV planner model to generate candidate behaviors and filters the data using rules and manual verification. Collision metrics, map-based metrics, and kinematic feasibility metrics are used to evaluate the safety and rule compliance of reactive responses. We construct 2,636 test scenarios with three categories and conduct a systematic evaluation of state-of-the-art models across multiple architectures, including Transformer-based, diffusion-based, and next-token-prediction-based models. We further analyze how replan frequency affects performance and provide insights for future studies.

Zhiyuan Zhang, Yanlun Peng, Jianing Zhang +7

Sch. of Computer Science & Sch. of Artificial Intelligence, Shanghai Jiao Tong University · Great Wall Motor · Institute of Trustworthy Embodied AI (TEAI), Fudan University +2

Autonomous Driving Benchmarks World Models for Autonomous Driving World Model Evaluation Autonomous Driving Simulation

Ask PDF

Oct 3, 2026

Human Behavior-Informed Crash Scenario Generation with Real-World Crash Priors for Autonomous Vehicle Safety Evaluation

Reliable safety evaluation of autonomous vehicles (AVs) is essential to improving road safety, yet it depends critically on realistic simulation of rare crashes. Existing crash scenario generation methods can increase collision occurrence, but often fail to realistically reproduce how crashes evolve before impact or the distribution of crash types observed in the real world. Here, we present CrashSim, a human behavior-informed crash scenario generation framework that uses real-world crash priors to guide generative multi-agent traffic simulation for more reliable AV safety evaluation. These priors capture how real-world crashes evolve before impact and how different crash types are distributed, allowing limited crash data to guide realistic and scalable scenario generation across naturalistic driving contexts. We evaluate CrashSim against competing methods, showing that it more closely reproduces real-world pre-impact behavior, collision dynamics, collision geometry and crash-type distributions. We further use CrashSim to construct nuCrash dataset, containing over 4,000 crash and near-crash scenarios. Closed-loop evaluation of five AV planners shows that nuCrash more effectively exposes differences in planner safety capabilities than nuScenes. An LLM-assisted evaluation agent further analyzes planner failures to provide capability-level diagnoses and targeted improvement guidance. Together, CrashSim enables realistic and scalable crash generation for more informative AV safety evaluation.

Mingxing Peng, Xusen Guo, Long Chen +3

The Hong Kong University of Science and Technology (Guangzhou), China · Institute of Automation Chinese Academy of Sciences, Beijing, China · The University of Hong Kong, Hong Kong, China +1

Autonomous Driving Scenario Generation Autonomous Driving Safety Evaluation Autonomous Driving Simulation Autonomous Driving Datasets

Ask PDF