Abstract:Accurately estimating time of arrival (ETA) for trucks is crucial for optimizing transportation efficiency in logistics. GPS trajectory data offers valuable information for ETA, but challenges arise due to temporal sparsity, variable sequence lengths, and the interdependencies among multiple trucks. To address these issues, we propose the Temporal-Attribute-Spatial Tri-space Coordination (TAS-TsC) framework, which leverages three feature spaces-temporal, attribute, and spatial-to enhance ETA. Our framework consists of a Temporal Learning Module (TLM) using state space models to capture temporal dependencies, an Attribute Extraction Module (AEM) that transforms sequential features into structured attribute embeddings, and a Spatial Fusion Module (SFM) that models the interactions among multiple trajectories using graph representation <a class="link-external link-http" href="http://learning.These" rel="external noopener nofollow">this http URL</a> modules collaboratively learn trajectory embeddings, which are then used by a Downstream Prediction Module (DPM) to estimate arrival times. We validate TAS-TsC on real truck trajectory datasets collected from Shenzhen, China, demonstrating its superior performance compared to existing methods.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to solve the problem of accurately predicting the truck arrival time (ETA, Estimated Time of Arrival) in logistics transportation. Specifically, the paper proposes a data - driven framework named TAS - TsC (Temporal - Attribute - Spatial Tri - space Coordination) to address the following three main challenges: 1. **Effective spatio - temporal trajectory data modeling**: - **Problem description**: The time - series features in GPS trajectory data are difficult to model effectively, especially when dealing with time - related features such as long - term dependencies, speed, and direction. - **Solution**: The Time Learning Module (TLM) in the TAS - TsC framework adopts a State Space Model (SSM), especially the Mamba model, to capture long - term dependencies and dynamic changes, thereby improving the accuracy of arrival time prediction. 2. **Data sparsity and variable sequence length**: - **Problem description**: Due to unstable GPS signals, device failures, and inconsistent data collection intervals, the trajectory data within the same range is sparse and the sequence lengths of different trajectories vary. - **Solution**: The TLM uses the Mamba model to capture long - term dependencies and update the hidden state at each time step; at the same time, the Attribute Extraction Module (AEM) extracts trajectory attributes through statistical methods, converting sparse and variable - length data into a unified representation that accurately reflects the trajectory attributes. 3. **Co - operation of multiple trajectories**: - **Problem description**: In actual logistics scenarios, the interaction between multiple truck trajectories significantly affects the estimation of arrival time. For example, at intersections or merging lanes, truck trajectories are not independent but interdependent. - **Solution**: The Spatial Fusion Module (SFM) constructs a spatio - temporal relationship graph to represent the interaction and spatial relationship between truck trajectories, and uses a graph diffusion method to integrate attributes and time features, capturing high - order structural information in the trajectories. ### Framework overview The TAS - TsC framework integrates three dimensions of time, attribute, and space, and works together through the following three modules: - **Time Learning Module (TLM)**: Adopts a State Space Model (SSM) to capture long - term dependencies and dynamic changes in time - series data. - **Attribute Extraction Module (AEM)**: Converts implicit trajectory features into structured attribute embeddings through feature engineering, enhancing the semantic richness of the features. - **Spatial Fusion Module (SFM)**: Constructs a spatio - temporal relationship graph, and uses a graph feature diffusion method to capture spatial embeddings, representing high - order structural information in the trajectories. Finally, the outputs of these modules are combined by the Downstream Prediction Module (DPM), using residual connections and Histogram - based Gradient Boosting (HGB) to accurately estimate the arrival time of the truck. ### Experimental verification The paper has been rigorously tested on real - world datasets, showing that the TAS - TsC framework is significantly superior to existing methods in arrival time prediction. ### Summary This paper proposes an innovative TAS - TsC framework, aiming to solve the difficult problem of predicting truck arrival time in logistics transportation by integrating three dimensions of time, attribute, and space.

TAS-TsC: A Data-Driven Framework for Estimating Time of Arrival Using Temporal-Attribute-Spatial Tri-space Coordination of Truck Trajectories

Adaptive Cross-platform Transportation Time Prediction for Logistics

Alleviating Data Sparsity Problems in Estimated Time of Arrival Via Auxiliary Metric Learning

Multi-View Spatial-Temporal Model for Travel Time Estimation

MDSTF: a multi-dimensional spatio-temporal feature fusion trajectory prediction model for autonomous driving

Route to Time and Time to Route: Travel Time Estimation from Sparse Trajectories

AttentionTTE: a deep learning model for estimated time of arrival

Deep spatial-temporal travel time prediction model based on trajectory feature

Automated data-driven prediction on aircraft Estimated Time of Arrival

TransETA: transformer networks for estimated time of arrival with local congestion representation

STC: Coarse-Grained Vehicular Data Based Travel Speed Sensing by Leveraging Spatial-Temporal Correlation

TITAN: A Spatiotemporal Feature Learning Framework for Traffic Incident Duration Prediction

Dynamic multiple-graph spatial-temporal synchronous aggregation framework for traffic prediction in intelligent transportation systems

SPTTE: A Spatiotemporal Probabilistic Framework for Travel Time Estimation

Prediction of Spatiotemporal Evolution of Urban Traffic Emissions Based on Taxi Trajectories

Aggressive Driving Saves More Time? Multi-task Learning for Customized Travel Time Estimation

TrajGAT: A map-embedded graph attention network for real-time vehicle trajectory imputation of roadside perception

Estimator: an Effective and Scalable Framework for Transportation Mode Classification over Trajectories

Characterized Diffusion and Spatial-Temporal Interaction Network for Trajectory Prediction in Autonomous Driving

A spatial–temporal graph-based AI model for truck loan default prediction using large-scale GPS trajectory data

TAML: A Traffic-aware Multi-task Learning Model for Estimating Travel Time