Multi-Context Fusion Transformer for Pedestrian Crossing Intention Prediction in Urban Environments
Researchers have developed a new AI model called the Multi-Context Fusion Transformer (MFT) to predict pedestrian crossing intentions in urban environments. The model combines multiple contextual attributes from four key dimensions - pedestrian behavior, environment, localization, and vehicle motion - to improve accuracy. Experimental results show MFT outperforms existing methods on three benchmark datasets, achieving accuracy rates of 73%, 93%, and 90%. The code for the mode
Researchers have developed a new AI model called the Multi-Context Fusion Transformer (MFT) to predict pedestrian crossing intentions in urban environments. The model combines multiple contextual attributes from four key dimensions - pedestrian behavior, environment, localization, and vehicle motion - to improve accuracy. Experimental results show MFT outperforms existing methods on three benchmark datasets, achieving accuracy rates of 73%, 93%, and 90%. The code for the model is available open-source.
---
Why it matters: This matters because accurate pedestrian intention prediction can significantly reduce traffic accidents and improve safety in urban areas. Engineers working on autonomous vehicles will be interested in this research as it provides a new approach to addressing a critical challenge in urban environments.
Source: https://arxiv.org/abs/2511.20011
This article was originally published at: https://arxiv.org/abs/2511.20011