Three-Step Performance Assessment of a Pedestrian Crossing Time Prediction Model
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 study addresses the challenge of calibrating pedestrian micro-simulation models, specifically focusing on validating a neural network-based predictive model for pedestrian crossing times. The research is motivated by the need for reliable simulation tools to study pedestrian safety and dynamics at conflict points, such as unsignalized crosswalks at roundabout entries. While micro-simulation software like Vissim/Viswalk is powerful, its accuracy depends on proper calibration of abstract parameters within the Social Force Model. The authors propose using a neural network to predict crossing times based on specific input parameters, serving as a tool to fine-tune the simulation model. The methodology employs a three-step validation procedure—visual, conceptual, and operational—applied to a case study of a roundabout in Monfalcone, Italy. The crosswalk, measuring 10.25 meters in length and 4 meters in width, was modeled in Vissim/Viswalk. The researchers selected eight input parameters for the neural network, comprising five pedestrian behavior parameters (e.g., relaxation time, anisotropy, social force strengths) and three vehicular car-following parameters (e.g., standstill distance, safety distance components). A training database of 100 parameter combinations was generated via simulation to train a ward neural network with three hidden layers. A separate prediction database, representing 20% of the training data, was used to test the network’s generalization capabilities. The results demonstrate a strong agreement between the neural network predictions and the micro-simulation outputs. Visual validation revealed a 97% correlation between predicted and simulated crossing times, with mean values of 6.32 seconds and 6.41 seconds, respectively. The mean absolute error for the training set was 0.559 seconds, while the generalization test yielded a 94% correlation and a mean absolute error of 1.76 seconds. Statistical analysis using the Kruskal-Wallis test indicated that all eight input parameters significantly influenced crossing time, with relaxation time, isotropic social force parameter, and the multiplicative part of safety distance being the most influential. Conceptual validation confirmed that the predicted and simulated times fell within ranges reported in literature for field measurements, despite the data exhibiting non-normal distribution as determined by the Anderson-Darling test. The significance of this work lies in establishing a robust framework for calibrating pedestrian micro-simulation models using neural networks. By validating the predictive model against simulation outputs and real-world literature ranges, the study confirms that neural networks can effectively replicate complex pedestrian behaviors governed by the Social Force Model. This approach facilitates the systematic calibration of simulation parameters, enhancing the reliability of transport infrastructure planning and safety assessments for vulnerable road users. The findings support the integration of artificial intelligence tools in transportation engineering to overcome the difficulties associated with measuring abstract simulation parameters.
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 author_sweep_intake on 2026-05-28.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | author_sweep | — | — | 2 | 2026-05-28 |
| archive | success | canonical_url | — | — | 7 | 2026-06-09 |
| extract | success | cached | — | — | 2 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-09 |
| chunk | success | chunk | — | — | 1 | 2026-06-09 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-09 |
| enrich | success | — | — | — | 1 | 2026-05-28 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 8 | 2026-06-11 |
| verify | success | — | — | — | 1 | 2026-06-10 |
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.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Theoretical Contribution: computational model