A novel definition of pedestrian-vehicle interaction as a partially observable Markov decision process (POMDP) with theory-grounded perceptual, cognitive, and motor constraints is introduced to establish a blueprint for simulator-ready pedestrian models that can support the development and evaluation of automated driving systems.
Abstract
Automated vehicles must be able to interact with pedestrians safely and efficiently across diverse traffic situations. Although driving simulators offer a scalable testbed for learning such capabilities, existing theory-inspired pedestrian models are narrow in scope and limited to go/no-go crossing decisions in single-lane settings. While data-driven approaches can predict pedestrian behavior in complex situations, they lack sufficient observations in rare, safety-critical scenarios. Here, we propose an approach to training pedestrian models in simulators so that learned policies generate demonstrably human-like behavior in realistic, complex traffic scenarios, including multiple lanes, heavy traffic, and dangerous driving styles. Our technical contribution is a novel definition of pedestrian-vehicle interaction as a partially observable Markov decision process (POMDP) with theory-grounded perceptual, cognitive, and motor constraints. It accounts for the highly adaptive nature of human behavior in traffic and simulates how people adjust their responses according to perceived danger, time pressure, and the complexity of the situation. When trained via deep reinforcement learning (RL) with domain randomization in a simulator, the model reproduces the broadest range of empirical findings shown so far on human crossing behavior, including gap acceptance, yielding acceptance, hesitation, and evasive speed adjustment. We show that learned policies transfer to unseen traffic environments, and can be further adapted to local traffic norms with finetuning. Together, these results establish a blueprint for simulator-ready pedestrian models that can support the development and evaluation of automated driving systems.
Autonomous driving systems inevitably operate in mixed traffic where autonomous vehicles coexist with humandriven vehicles (HDVs). Motion planning must therefore handle uncertain intent, heterogeneous driving styles, asymmetric responsibility, and behavioral adaptation to the ego vehicle’s actions. This survey reviews...
Bai Li, Jin-Di Hao, Xiao-Han Yang et al.· IEEE/ASME International Conf...· 0 citations
As the microscopic foundation of traffic dynamics, driving interactions and behaviors are critical for understanding the operational mechanisms of traffic systems and enabling safe, human-like autonomous driving. However, most existing physics-based driving behavior models are suitable for specific scenarios, and data-...
Jing-Hua Wang, Guang-Quan Lu, Miao-Miao Liu et al.· Accident Analysis and Preven...· 0 citations
An expert-guided framework to quantify pedestrian-vehicle interaction risk by learning the takeover behaviors of safety drivers in AVs is developed and validates and models interaction risk based on experienced drivers' risk perception and highlights key factors influencing human risk assessment.
Yiran Luo, Quan Li, Si-Yuan Liu et al.· Traffic Injury Prevention· 0 citations
The proposed Interaction-Field Residual Learning for Calibrated Lane-Change Intent framework introduces a continuous interaction-field representation that models traffic participants as sources of physical potentials, thereby capturing both repulsive and attractive social forces in a unified topographical space.
Anele Mthembu, Thabo Pretorius· Journal of innovative resear...· 0 citations
This study considers a merging scenario on a congested highway on-ramp and develops a system that autonomously performs lane changes from an on-ramp to the main lane and constructs a method for selecting a target space in a congested traffic flow into which the host vehicle enters itself.
Hanwool Woo, Keiju Nishimura, Takumi Iwasa· Journal of Robotics and Mech...· 0 citations
The results showed that ATs increased perceived risk and encouraged more cautious crossing behavior, suggesting a risk-compensation effect, but this compensation was weakened under rainy conditions, where braking-related safety margins were reduced.