Towards An Efficient and Effective En Route Travel Time Estimation Framework

TL;DR

This work tackles en-route travel time estimation (ER-TTE) by addressing the real-time efficiency bottleneck caused by frequent re-estimation under high request loads. It introduces U-ERTTE, a framework that combines Uncertainty-Guided Decision (UGD) to selectively trigger re-estimation using confidence intervals, with Fine-Tuning with Meta-Learning (FTML) to learn general driving patterns and rapidly adapt to route-specific conditions via meta-learning. The approach yields substantial gains in inference speed and throughput while preserving high accuracy, as demonstrated on two large real-world datasets (Porto and Xian) with comprehensive ablations and online tests. The work advances ER-TTE by delivering a practical, scalable solution that balances responsiveness and predictive quality, with code available for reproducibility.

Abstract

En route travel time estimation (ER-TTE) focuses on predicting the travel time of the remaining route. Existing ER-TTE methods always make re-estimation which significantly hinders real-time performance, especially when faced with the computational demands of simultaneous user requests. This results in delays and reduced responsiveness in ER-TTE services. We propose a general efficient framework U-ERTTE combining an Uncertainty-Guided Decision mechanism (UGD) and Fine-Tuning with Meta-Learning (FTML) to address these challenges. UGD quantifies the uncertainty and provides confidence intervals for the entire route. It selectively re-estimates only when the actual travel time deviates from the predicted confidence intervals, thereby optimizing the efficiency of ER-TTE. To ensure the accuracy of confidence intervals and accurate predictions that need to re-estimate, FTML is employed to train the model, enabling it to learn general driving patterns and specific features to adapt to specific tasks. Extensive experiments on two large-scale real datasets demonstrate that the U-ERTTE framework significantly enhances inference speed and throughput while maintaining high effectiveness. Our code is available at https://github.com/shenzekai/U-ERTTE

Figures (7)

• Figure 1: Uncertainty quantification in ER-TTE using confidence intervals to determine when re-estimation is necessary.
• Figure 2: An explanation of Meta-Learning for ER-TTE
• Figure 3: Pipeline of our U-ERTTE framework.
• Figure 4: Components of the U-ERTTE Framework
• Figure 5: The efficiency comparison of inference time and throughput.