Abstract:With the development of location-acquisition technologies, there are a huge number of mobile trajectories generated and accumulated in a variety of domains. However, due to the constraints of device and environment, many trajectories are recorded at low sampling rate, which increases the uncertainty between two consecutive sampled points in the trajectories. Our task is to recover a high-sampled trajectory based on the irregular low-sampled trajectory in free space, i.e., without road network information. There are two major problems with traditional solutions. First, many of these methods rely on heuristic search algorithms or simple probabilistic models. They cannot well capture complex sequential dependencies or global data correlations. Second, for reducing the predictive complexity of the unconstrained numerical coordinates, most of the previous studies have adopted a common preprocessing strategy by mapping the space into discrete units. As a side effect, using discrete units is likely to bring noise or inaccurate information. Hence, a principled post-calibration step is required to produce accurate results, which has been seldom studied by existing methods. To address the above difficulties, we propose a novel Deep Hybrid Trajectory Recovery model, named DHTR. Our recovery model extends the classic sequence-to-sequence generation framework by implementing a subsequence-to-sequence recovery model tailored for the current task, named subseq2seq. In order to effectively capture spatiotemporal correlations, we adopt both spatial and temporal attentions for enhancing the model performance. With the attention mechanisms, our model is able to characterize long-range correlations among trajectory points. Furthermore, we integrate the subseq2seq with a calibration component of Kalman filter (KF) for reducing the predictive uncertainty. At each timestep, the noisy predictions from the subseq2seq component will be fed into the KF component for calibration, and then the refined predictions will be forwarded to the subseq2seq component for the computation of the next timestep. Extensive results on real-world datasets have shown the superiority of the proposed model in both performance and interpretability.

MTrajRec: Map-Constrained Trajectory Recovery via Seq2Seq Multi-task Learning

MTrajRec

Modeling Trajectories with Multi-task Learning

Trajectory-as-a-Sequence: A Novel Travel Mode Identification Framework

RNTrajRec: Road Network Enhanced Trajectory Recovery with Spatial-Temporal Transformer

Mining Urban Moving Tra jectory Patterns Based on Multi-scale Space Partition and Road Network Modeling

MSTM: A novel map matching approach for low-sampling-rate trajectories

Learning to Generate Maps from Trajectories

TrajRecovery: an Efficient Vehicle Trajectory Recovery Framework Based on Urban-Scale Traffic Camera Records

A Critical Perceptual Pre-trained Model for Complex Trajectory Recovery

Towards Effective Transportation Mode-Aware Trajectory Recovery: Heterogeneity, Personalization and Efficiency

Robust Dynamic Trajectory Regression on Road Networks: A Multi-task Learning Framework

Travel Mode Identification for Non-Uniform Passive Mobile Phone Data

Large-scale Joint Map Matching of GPS Traces

Surrogate Modeling of Trajectory Map-matching in Urban Road Networks using Transformer Sequence-to-Sequence Model

More Than Routing: Joint GPS and Route Modeling for Refine Trajectory Representation Learning

Micro-Macro Spatial-Temporal Graph-based Encoder-Decoder for Map-Constrained Trajectory Recovery

High-resolution Reconstruction of Non-Motorized Trajectory in Shared Space: A New Approach Integrating the Social Force Model and Particle Filtering

MFTraj: Map-Free, Behavior-Driven Trajectory Prediction for Autonomous Driving

Learning Universal Trajectory Representation via a Siamese Geography-Aware Transformer

Deep Trajectory Recovery with Fine-Grained Calibration Using Kalman Filter