Understanding and Guiding Pedestrian and Crowd Motion

Yang, Dongfang; Redmill, Keith; Ozguner, Umit · 2020 · ROSA P / Mobility21, Carnegie Mellon University

archive: archived pipeline: cataloged verified

Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)

Summary

This research report addresses the challenge of modeling pedestrian motion in shared spaces where automated vehicles and pedestrians interact, a critical component for developing safe first-mile and last-mile transportation systems. Motivated by rising pedestrian fatalities and the need for intelligent systems to predict crowd behavior in environments like university campuses, the authors aim to create a mathematical model that accurately simulates how pedestrians react to both surrounding crowds and low-speed vehicles. The work is grounded in the Ohio State University’s SMOOTH initiative, which utilizes automated shuttles in pedestrian-heavy zones. The study employs a modified social force model, treating pedestrians as point-mass agents whose motion is governed by Newtonian dynamics. The total force acting on a pedestrian is the summation of destination attraction, repulsion from other pedestrians, and a newly designed vehicle effect. The authors developed three iterations of the vehicle effect, culminating in a model that accounts for anisotropy and specific spatial zones (front, body, and rear) relative to the vehicle. To validate and calibrate this model, the researchers created two datasets, CITR and DUT, capturing vehicle and pedestrian trajectories. They extracted these trajectories using video stabilization, tracking algorithms, and coordinate transformation. Model calibration was performed using a genetic algorithm to optimize parameters against the empirical data, ensuring the model could replicate fundamental interaction scenarios involving at least five pedestrians and vehicles moving under 4 m/s. The results demonstrate that the calibrated social force model effectively reproduces realistic pedestrian-pedestrian and vehicle-pedestrian interactions. The model successfully captures complex behaviors such as lane formation, collision avoidance, and the varying influence of a vehicle based on its position and speed relative to pedestrians. Specifically, the inclusion of anisotropic functions and decaying force magnitudes allowed the model to distinguish between interactions from different angles and distances, improving prediction accuracy. The calibrated parameters showed that pedestrians adjust their velocity and acceleration limits based on crowd density and the magnitude of vehicle influence, switching from destination-oriented movement to avoidance behaviors when vehicle proximity increases. The significance of this work lies in providing a robust framework for integrating pedestrian behavior prediction into automated vehicle control systems. By accurately simulating multi-pedestrian interactions with vehicles, the model enables automated vehicles to make more reliable local reactive decisions and path planning choices in shared spaces. This contributes to enhancing transportation efficiency and safety in hybrid traffic environments, supporting the deployment of on-demand automated vehicles in areas where traditional lane-based traffic rules do not apply. The report concludes by outlining applications for pedestrian detection and scenario prediction, highlighting the model's potential to improve the integration of automated mobility solutions into existing urban infrastructure.

Key finding

A calibrated social force model using a genetic algorithm successfully reproduces fundamental vehicle-pedestrian interaction trajectories by integrating anisotropic vehicle effects and dynamic velocity constraints.

Methodology

modeling

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• pedestrian behavior perception