MixTTE: Multi-Level Mixture-of-Experts for Scalable and Adaptive Travel Time Estimation

2026-01-06

arXiv.org

AI Summary

This work addresses the limitations of existing travel time estimation methods in effectively modeling city-scale traffic dynamics and long-tail scenarios, which often result in suboptimal prediction performance across large-scale road networks. To overcome these challenges, we propose a scalable and adaptive travel time estimation framework that jointly captures local segment-level dependencies and global route-level dynamics through a hierarchical mixture-of-experts architecture. Key innovations include a spatiotemporal external attention mechanism to model cross-regional temporal correlations, a graph-structured stable Mixture-of-Experts network, and an asynchronous incremental learning strategy for efficient online updates. Evaluated on real-world large-scale datasets, our approach significantly outperforms seven strong baselines and has been deployed on the DiDi platform, substantially improving both estimation accuracy and system stability.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Accurate Travel Time Estimation (TTE) is critical for ride-hailing platforms, where errors directly impact user experience and operational efficiency. While existing production systems excel at holistic route-level dependency modeling, they struggle to capture city-scale traffic dynamics and long-tail scenarios, leading to unreliable predictions in large urban networks. In this paper, we propose \model, a scalable and adaptive framework that synergistically integrates link-level modeling with industrial route-level TTE systems. Specifically, we propose a spatio-temporal external attention module to capture global traffic dynamic dependencies across million-scale road networks efficiently. Moreover, we construct a stabilized graph mixture-of-experts network to handle heterogeneous traffic patterns while maintaining inference efficiency. Furthermore, an asynchronous incremental learning strategy is tailored to enable real-time and stable adaptation to dynamic traffic distribution shifts. Experiments on real-world datasets validate MixTTE significantly reduces prediction errors compared to seven baselines. MixTTE has been deployed in DiDi, substantially improving the accuracy and stability of the TTE service.

Problem

• Travel Time Estimation
• traffic dynamics
• long-tail scenarios
• large urban networks
• prediction reliability

Innovation

• Mixture-of-Experts
• Travel Time Estimation
• Spatio-temporal Attention
• Graph Neural Networks
• Incremental Learning