RLOMM: An Efficient and Robust Online Map Matching Framework with Reinforcement Learning

Ao Zhou; Haitao Yuan; Minxiao Chen; Nan Jiang; Sai Wu; Shangguang Wang; Zhihan Zheng

REVIEW 4 major objections 5 minor 1 cited by

RLOMM: An Efficient and Robust Online Map Matching Framework with Reinforcement Learning

T0 review · 4 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that RLOMM, a deep reinforcement learning framework for online map matching, outperforms existing rule-based and deep-learning methods in accuracy, efficiency, and robustness on three real-world datasets.

desk verdict A coherent RL framework for online map matching with real efficiency gains, but the accuracy margins on two of three datasets rest on HMM-generated labels and no code or error bars. read the letter →

arxiv 2502.06825 v2 pith:7Y7PFUU3 submitted 2025-02-05 cs.LG cs.DB

Minxiao Chen , Haitao Yuan , Nan Jiang , Zhihan Zheng , Sai Wu , Ao Zhou , Shangguang Wang This is my paper · ORCID

classification cs.LGcs.DB

keywords mapmatchingonlinereinforcementlearningdeepQ-learningtrajectoryroadnetworkgraphneuralcontrastive

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper is trying to establish that online map matching—the task of snapping a live GPS stream to road segments as points arrive—does not have to be an expensive replay of offline matching over growing prefixes. Its proposed framework, RLOMM, casts the problem as an Online Markov Decision Process, learns decisions with deep Q-learning and a future-oriented reward, and uses separate graph encoders plus contrastive alignment to fuse trajectory and road information. The authors report that on three real-world taxi datasets and at three sampling rates, RLOMM outperforms existing rule-based and deep-learning methods in trajectory-level accuracy and longest-common-subsequence ratio, while also reducing matching time to about one second per 10K trajectories and cutting memory usage relative to other learned methods. If this holds, a service provider could use coarser GPS sampling and still get better matching than current methods at the finer rate, lowering cost and latency simultaneously.

What carries the argument

The load-bearing mechanism is the OMDP-plus-RL loop. At each online step, the state is assembled from real-time components (the current trajectory points, previously matched roads, candidate roads) and historical components (the previous RNN hidden states for trajectory and road), so no full historical replay is needed. The action selects one candidate road per incoming point using attention scores over a fusion of trajectory and road embeddings. The reward function is the second pillar: it rewards exact matches, rewards sustained runs of correct matches, penalizes detours that back-track over recently matched roads, and rewards choosing roads that are well-connected to the previous match, which makes the agent plan ahead. The third pillar is the dual-graph encoder: a trajectory transition graph (grid cells as nodes, transition counts as edge weights) encoded by a graph convolutional network, and a link connection graph (road segments as nodes) encoded by a graph isomorphism network, with a mapping matrix that seeds trajectory node features from road representations. An InfoNCE contrastive loss aligns the trajectory anchor representation with the ground-truth road segment against negative candidates. Training uses Double DQN with a Huber temporal-difference loss together with the contrastive loss.

What would settle it

Take a held-out sample of Porto and Chengdu trajectories, have annotators or an independent high-accuracy system determine the true road for each point, and recompute AccT and LCSR for RLOMM and the best baseline on that sample; if the accuracy gap shrinks or reverses, the central claim that RLOMM fundamentally outperforms existing methods would be called into question.

Watch

Extended reading notes

Core claim

The paper's central discovery is that online map matching—snapping each incoming GPS point to a road segment as it arrives—is better modeled as an Online Markov Decision Process than as a repeated invocation of offline matching. In RLOMM, the state at each step carries the current batch of trajectory points, the previously matched roads, the set of candidate road segments, and the compressed historical context held in the encoding networks' hidden states. The policy is learned with Double DQN, with a reward that combines immediate match accuracy, a bonus for consecutive successes, a penalty for detours, and a term for road connectivity, so the agent optimizes a sequence of matches rather than a single greedy choice. To bridge the heterogeneity between trajectories and road networks, the model builds a trajectory transition graph over grid cells and a link connection graph over road segments, encodes them with GCN and GIN respectively, and aligns the two modalities in latent space with an InfoNCE contrastive loss. The reported experiments show AccT up to 77.78% on Beijing at 50% sampling—about 12 points above the strongest baseline—with matching time around one second per 10K trajectories and substantially lower memory usage than other learning-based methods.

Load-bearing premise

The ground-truth labels for two of the three datasets were produced by an HMM annotator at the original sampling rate, and the paper assumes those labels are accurate enough to train and evaluate against—even though HMM is also one of the baselines, so any systematic labeling error could inflate the reported advantage.

Editorial extensions

If this is right

Operators can drop GPS sampling from 15 seconds to 30, 60, or 120 seconds and still match trajectories more accurately than the best baseline at the original rate, cutting data volume and cost.
Online inference on 10K trajectories takes about one second, making the method practical for real-time navigation and fleet tracking at city scale.
The reward's consecutive-success, detour-penalty, and road-connectivity terms reduce cascading errors, so a wrong match is less likely to corrupt later decisions.
The same graph encoders and contrastive alignment handle both dense and sparse trajectories; the reported gains at 12.5% sampling show the design degrades gracefully as data sparsifies.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Because the Porto and Chengdu labels come from an HMM annotator, part of the reported gap over HMM-based baselines may reflect label bias; a human-verified subset would settle how much of the gain is real.
The reward weights α, β, γ are fixed by hand per dataset; a learned or adaptive reward schedule could plausibly close the remaining gap to perfect matching across cities with very different road densities.
The OMDP formulation is a general recipe for streaming decision problems where history must be compressed rather than replayed; online route recovery and real-time vessel tracking are natural testbeds.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

Read RLOMM. It deserves a serious referee, and if the numbers hold up it would be a practically useful contribution. The core idea is genuinely new for online map matching: modeling the problem as an Online MDP, keeping the RNN hidden states as state compression, and using two graph encoders (GIN for the road network, GCN for a trajectory transition graph) with a contrastive alignment loss. The efficiency argument is sound: you avoid re-encoding the whole history at every step, and the complexity table makes that concrete. The ablations (NC, NI, NM) show the graph components matter, and the hyperparameter sensitivity figures are a nice touch.

The soft spots are all on the empirical side. The main one is the ground truth. For Porto and Chengdu the labels are generated by HMM at 15s sampling, and HMM is also a baseline. That creates a real risk that part of the reported gap against HMM/FMM/AMM on those two datasets is label bias rather than matching quality. It is not fatal: Beijing has real labels and the gains are consistent there, so the effect, if any, is likely inflating the margin rather than creating it. But it needs to be stated more carefully, and the authors should provide error bars or at least multiple seeds. Right now Table 3 is a single run with no variance, so a skeptical reader cannot check whether 77.78% vs 66.01% is stable. No code release makes that worse.

Two smaller things. First, the RL component itself is not ablated: the ablations change the encoders, not the learning objective. A supervised version of the same architecture would tell you whether the future-aware reward actually buys anything beyond the encoders. The case study hints that the specific rewards help, but it is not systematic. Second, the complexity analysis is fair for the matching loop but omits the cost of building and updating the trajectory transition graph; for a system paper that matters at deployment scale.

Overall: the paper is coherent, the method is plausible, and the efficiency gains (3x over AMM) look credible even without the accuracy claims. The citation pattern is fine. The HMM-label issue is the kind of thing that should be fixed in revision, not a reason to reject.

Recommendation: send to peer review. The work is new enough and the evaluation is thorough enough to merit referee time; just flag the label question and the missing variance.

Referee Report

4 major / 5 minor

Summary. The paper proposes RLOMM, an online map matching framework that models the problem as an Online Markov Decision Process (OMDP) and solves it with deep Q-learning. The framework pairs separate graph encoders for a trajectory transition graph and a road link connection graph, uses RNNs to carry historical information across matching steps, and adds a contrastive trajectory-road alignment loss plus a four-term reward (accuracy, consecutive success, detour penalty, road connectivity). The evaluation on Beijing, Porto, and Chengdu at three sampling rates compares RLOMM with MDP, HMM, FMM, AMM, MTrajRec, L2MM, and GraphMM, reporting consistently higher AccT and LCSR, better efficiency, and additional ablations, scalability experiments, and a case study.

Significance. If the empirical claims hold, the paper is a solid contribution to online map matching: it addresses a real operational need, the architecture is coherent, and the evaluation is unusually broad, covering three datasets, multiple sampling rates, seven baselines, ablations, scalability, and a case study. The reported efficiency gains are large, and the case study gives qualitative evidence that the designed reward terms do what they are supposed to do. The main reservations are external validity rather than internal inconsistency: two of the three datasets use HMM-generated labels while HMM-family methods are also the baselines; no variance or significance information is reported for the central accuracy tables; and the complexity analysis does not fully account for the road-graph encoding. These issues are fixable with additional experiments and clarifications, so the paper should be revised rather than rejected.

major comments (4)

[Section 6.1 (Data Labels) and Table 3] The ground-truth labels for Porto and Chengdu are generated by HMM annotation at 15s sampling, and HMM, FMM, and AMM (all HMM-based) are among the baselines. Because the reward (Eqs. 13-15) and the contrastive loss (Eq. 10) train directly against these labels, and because AccT/LCSR in Table 3 are computed against the same label source, a systematic bias in the HMM annotations could inflate RLOMM's advantage over HMM-family baselines on exactly the two datasets that are not Beijing. The statement that HMM is effective at high sampling rates is a consensus point, but it does not validate the labels at the evaluation frequencies used here, nor does it rule out a shared design between the annotation HMM and the baseline HMM/FMM/AMM implementations. Please add an independent label-quality check (for example, manual inspection of a sampled subset, or comparison with real labels on a subset), state whether the annotation HMM is the same implementation as the baseline HMM, and release the annotation code. If such validation is not possible, the strong accuracy claims should be explicitly scoped to the Beijing dataset, whose labels come from Tencent Maps.
[Table 3 and Section 6.4] The central claim that RLOMM 'significantly outperforms' existing methods is supported only by point estimates of AccT and LCSR. No standard deviations, confidence intervals, number of test trajectories, or significance tests are reported for any entry of Table 3. Some margins are large (for example, Beijing 50% AccT: 77.78% vs. 66.01% for GraphMM), so the point estimates are suggestive, but the word 'significantly' in the abstract and Section 6.4 requires statistical support. Please report means and variances over multiple training runs, or at least paired tests or confidence intervals, for the main comparison table.
[Section 5.3, Table 1, and Section 4.1] The complexity analysis states that RLOMM has parameter complexity O(d^2) and time complexity O(n_T * l_T/a * d^2), with the justification that the method 'does not require embedding encoding of n_R road segments.' However, Section 4.1 defines road representations by applying GIN over the full link connection graph G_R (Eq. 1), whose node set is the set of all n_R road segments, and candidate embeddings are selected from the global representation z_r^G. If z_r^G is precomputed offline once per graph, this must be stated explicitly, the precomputation cost must be reported separately, and the matching-time comparison in Table 4 should clearly exclude only that shared precomputation. If z_r^G is recomputed during online matching, the missing O(|E_R| d^2) graph-convolution cost makes the table's claimed online complexity incomplete. Please clarify this point, since the efficiency claim is load-bearing for the paper's practical contribution.
[Section 5.2, Algorithm 1, and Section 3] The reward function requires the ground-truth road segment y_si at every step (Eqs. 13-15), and the contrastive anchor-positive construction also uses y_i (Eq. 10). This is a supervised training signal, not an environment reward that would be available at deployment time. The abstract and Section 3 describe the method as 'continuously update and optimize during the decision-making process based on feedback,' but the evaluated system appears to be trained offline on labeled data and then deployed. Please state clearly whether online fine-tuning is assumed, and if so, how rewards are obtained when no ground-truth labels are available in a live streaming setting. This distinction matters for the claimed robustness and online adaptability of the framework.

minor comments (5)

[Section 6.3] The phrase 'Based on the majority voting rule' is unexplained; please specify how the best lambda value was selected across the three datasets, since Figure 6 appears to show per-dataset curves.
[Section 4.2 (Remark)] The claim that RNN is preferable to GRU, LSTM, and Transformer for this task is argued by design intuition but not tested empirically; a sentence acknowledging the absence of a sequence-encoder ablation would be appropriate.
[Section 2.2 and Eq. (12)] The notation a_i is used for the vector of k actions in the action definition and also for the individual time-step action a_i^(n) in Eq. (12); please define this indexing explicitly when the action is first introduced.
[Algorithm 1] The pseudo-code invokes ModelTrain inside the mini-batch loop (line 13) but does not show where the batch size b_s is used inside ModelTrain, and the update of the main network on line 14 appears to happen before the target-network update on lines 15-16; please clarify the control flow.
[Figures 5-7] Please ensure that all subfigures have readable axis labels, legends, and marked optimal values; in the current reproduction, several panels are difficult to interpret without the accompanying text.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found; the method is self-contained and evaluated against external benchmarks, with only an HMM-label validity risk that is not a derivation loop.

full rationale

The paper's derivation chain is self-contained. The OMDP formulation, graph encoders, contrastive alignment loss, and DQN-style training objective are all defined from primitive quantities (states, actions, rewards, ground-truth labels) rather than from the paper's own conclusions. The reward function uses ground truth during training, but this is standard supervised reward design, not a prediction that is later reported as an empirical result. At inference, the arg-max over candidate scores (Eq. 12) does not depend on ground truth, and the reported AccT/LCSR numbers are computed on held-out test sets against external baselines (MDP, HMM, FMM, AMM, MTrajRec, L2MM, GraphMM). The self-citations in the reference list (e.g., the authors' prior CIKM, ICDE, and SIGMOD papers) are to related spatiotemporal work and are not load-bearing for the central map-matching claim; no uniqueness theorem or ansatz is imported from same-author prior work. The one legitimate concern is the paper's own statement in Section 6.1, 'Data Labels,' that Porto and Chengdu labels are generated by HMM annotation at 15s sampling, while HMM-based methods are also baselines. This is a label-validity and potential benchmark-bias risk, because training and evaluation share the same label source; however, it is not circularity in the derivation: RLOMM's outputs are not constructed to equal HMM labels by definition, and the Beijing dataset uses real-world Tencent Maps labels with the same qualitative pattern of reported gains. Therefore no circular step satisfies the evidentiary standard of exhibiting an equation or fitted parameter that reduces to its own input, and the appropriate score is 0.

Assumptions & free parameters 15 free parameters · 5 assumptions · 0 invented entities

The central claims rest on the validity of HMM-generated labels, static road networks, candidate set completeness, grid mapping, and convergence of the RL training; these are domain assumptions not proven by the paper. The free parameters are hyperparameters tuned on validation data or chosen ad hoc, affecting the reported performance.

free parameters (15)

Learning rate = 0.001
Set in experimental settings; standard Adam learning rate, not tuned in the paper.
Batch size = 512
Set in experimental settings; not tuned in the paper.
Training epochs = 200
Set with early stopping; not tuned in the paper.
RNN layers for trajectory N_t = Beijing:3, Porto:2, Chengdu:3
Tuned on validation sets; see Section 6.2.
RNN layers for road N_r = Beijing:3, Porto:4, Chengdu:4
Tuned on validation sets; see Section 6.2.
GCN layers N_c = Beijing:2, Porto:3, Chengdu:3
Tuned on validation sets; see Section 6.2.
GIN layers N_i = Beijing:3, Porto:4, Chengdu:3
Tuned on validation sets; see Section 6.2.
RNN hidden dimension d_h = Beijing:128, Porto:64, Chengdu:64
Tuned on validation sets; see Section 6.2.
Attention dimension d_a = Beijing:128, Porto:16, Chengdu:32
Tuned on validation sets; see Section 6.2.
Loss weight lambda = 0.1
Tuned from 0 to 1 on validation data; see Section 6.3.
Grid side length l_g = Beijing:5m, Porto:5m, Chengdu:10m
Tuned on validation data; see Section 6.3.
Reward weight alpha = 0.01
Chosen for consecutive success reward; no tuning described.
Reward weight beta = 0.05
Chosen for detour penalty; no tuning described.
Reward weight gamma = 0.02
Chosen for road connectivity reward; no tuning described.
Number of candidates n_c = 10
Set to 10, covering nearly all road segments within 140-175m of the trajectory points.

assumptions (5)

domain assumption HMM-generated labels for Porto and Chengdu are reliable ground truth for training and evaluation.
Section 6.1 states labels come from HMM annotation under high sampling rates, a common practice, but this assumption is load-bearing because HMM is also used as a baseline.
domain assumption The road network is fixed and does not include dynamic changes such as temporary road closures.
Section 5.2 explicitly states this, limiting the robustness evaluation to static networks.
domain assumption Precomputed candidate sets based on spatial distance always contain the true road segment.
Problem definition in Section 2.2 uses candidate sets computed by spatial distance; with n_c=10, the paper argues it covers nearly all roads within 160m/140m/175m, but this is an empirical claim not proven.
domain assumption Grid representation of trajectories preserves enough information for accurate matching.
Section 2.1 maps trajectory points to grids, which aggregates points and may lose precision at grid boundaries.
domain assumption The reinforcement learning training procedure with the designed reward converges to a good policy.
The paper uses Double DQN and reports good results, but no convergence analysis or variance across runs is provided.

how reviews work

0 comments

Cite this review

Pith. "Pith review of RLOMM: An Efficient and Robust Online Map Matching Framework with Reinforcement Learning." pith.science (2026). https://pith.science/paper/7Y7PFUU3

@misc{pith2026250206825,
  author       = {Pith},
  title        = {Pith review of: RLOMM: An Efficient and Robust Online Map Matching Framework with Reinforcement Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7Y7PFUU3}},
  note         = {Machine review of arXiv:2502.06825}
}

read the original abstract

Online map matching is a fundamental problem in location-based services, aiming to incrementally match trajectory data step-by-step onto a road network. However, existing methods fail to meet the needs for efficiency, robustness, and accuracy required by large-scale online applications, making this task still challenging. This paper introduces a novel framework that achieves high accuracy and efficient matching while ensuring robustness in handling diverse scenarios. To improve efficiency, we begin by modeling the online map matching problem as an Online Markov Decision Process (OMDP) based on its inherent characteristics. This approach helps efficiently merge historical and real-time data, reducing unnecessary calculations. Next, to enhance robustness, we design a reinforcement learning method, enabling robust handling of real-time data from dynamically changing environments. In particular, we propose a novel model learning process and a comprehensive reward function, allowing the model to make reasonable current matches from a future-oriented perspective, and to continuously update and optimize during the decision-making process based on feedback. Lastly, to address the heterogeneity between trajectories and roads, we design distinct graph structures, facilitating efficient representation learning through graph and recurrent neural networks. To further align trajectory and road data, we introduce contrastive learning to decrease their distance in the latent space, thereby promoting effective integration of the two. Extensive evaluations on three real-world datasets confirm that our method significantly outperforms existing state-of-the-art solutions in terms of accuracy, efficiency and robustness.

Figures

Figures reproduced from arXiv: 2502.06825 by the authors.

**Figure 2.** An illustration of Grids Partition, Trajectory Transition Graph, and Link Connection Graph. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** The architecture of RLOMM, which consists of two parts: Online Inference and Offline Training. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: The framework of feature encoding. set of neighbor nodes of𝑢𝑗 , and 𝑧 (𝑛−1) 𝑢𝑗 and 𝑧 (𝑛−1) 𝑣 are the feature vectors of node𝑢𝑗 and its neighbors, respectively. For previously matched road segments 𝑈𝑖 = 𝑈𝑗:𝑗+𝑘 = ⟨𝑢𝑗 , 𝑢𝑗+1, . . . , 𝑢𝑗+𝑘−1⟩, we use the feature ⟨𝑧 (𝑛) 𝑢𝑗 …

**Figure 5.** Figure 5: AccT & LCSR vs. Hyper-parameters scenarios. Note that it’s fair for them, as these offline methods obtain all the currently available trajectories each time a match is made, which is consistent with the setting of offline matching problem they aim to solve. Evaluation …

**Figure 6.** Figure 6: AccT & LCSR vs. the loss weight 𝜆 and the grids side length 𝑙𝑔 𝑁𝑟 = 4, 𝑁𝑐 = 3, 𝑁𝑖 = 4, 𝑑ℎ = 64, 𝑑𝑎 = 16. (3) For Chengdu, we have 𝑁𝑡 = 3, 𝑁𝑟 = 4, 𝑁𝑐 = 3, 𝑁𝑖 = 3, 𝑑ℎ = 64, 𝑑𝑎 = 32. 6.3 Effectiveness of Loss Weight and Partition To fine-tune the loss weight 𝜆, we vary it…

**Figure 7.** Figure 7: AccT & LCSR vs. the Scalability. we apply parallel mechanisms to ensure fairness compared to learning-based methods running on the GPU with batch. The outcomes of these evaluations are presented in [PITH_FULL_IMAGE:figures/full_fig_p021_7.png]

**Figure 8.** Figure 8: An illustration of matching result for two trajectories of Porto. The blue points are the trajectory points [PITH_FULL_IMAGE:figures/full_fig_p022_8.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

EDPC: Accelerating Lossless Compression via Lightweight Probability Models and Decoupled Parallel Dataflow
cs.IT 2025-07 conditional novelty 4.0 of 10

EDPC reports state-of-the-art lossless compression ratios across text, image, float, and benchmark datasets while running about 2.7x faster than the strongest prior learned compressor, PAC.

Reference graph

Works this paper leans on

55 extracted references · 49 canonical work pages · cited by 1 Pith paper

[1]

2015. kaggle. https://www.kaggle.com/c/pkdd-15-taxi-trip-time-prediction-ii

work page 2015
[2]

2021. GAIA. https://outreach.didichuxing.com/research/opendata/

work page 2021
[3]

Pingfu Chao, Yehong Xu, Wen Hua, and Xiaofang Zhou. 2020. A Survey on Map-Matching Algorithms. InADC, Renata Borovica-Gajic, Jianzhong Qi, and Weiqing Wang (Eds.), Vol. 12008. Springer, 121–133

work page 2020
[4]

Minxiao Chen, Haitao Yuan, Nan Jiang, Zhifeng Bao, and Shangguang Wang. 2024. Urban Traffic Accident Risk Prediction Revisited: Regionality, Proximity, Similarity and Sparsity. InCIKM. ACM, 281–290

work page 2024
[5]

Kyunghyun Cho, Bart van Merrienboer, Çaglar Gülçehre, Dzmitry Bahdanau, Fethi Bougares, Holger Schwenk, and Yoshua Bengio. 2014. Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation. In EMNLP, Alessandro Moschitti, Bo Pang, and Walter Daelemans (Eds.). ACL, 1724–1734

work page 2014
[6]

Sean R Eddy. 1996. Hidden markov models. In Current opinion in structural biology , Vol. 6. Elsevier, 361–365

work page 1996
[7]

Jie Feng, Yong Li, Kai Zhao, Zhao Xu, Tong Xia, Jinglin Zhang, and Depeng Jin. 2022. DeepMM: Deep Learning Based Map Matching With Data Augmentation. In TMC, Vol. 21. 2372–2384

work page 2022
[8]

Xiao Fu, Jiaxu Zhang, and Yue Zhang. 2021. An Online Map Matching Algorithm Based on Second-Order Hidden Markov Model. In JAT, Vol. 2021. 9993860

work page 2021

Show all 55 references

[9]

Chong Yang Goh, Justin Dauwels, Nikola Mitrovic, Muhammad Tayyab Asif, Ali Oran, and Patrick Jaillet. 2012. Online map-matching based on Hidden Markov model for real-time traffic sensing applications. In ITSC. IEEE, 776–781

2012
[10]

Chenjuan Guo, Bin Yang, Jilin Hu, and Christian S. Jensen. 2018. Learning to Route with Sparse Trajectory Sets. In ICDE. IEEE Computer Society, 1073–1084

2018
[11]

Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long Short-Term Memory. In Neural Comput., Vol. 9. 1735–1780

1997
[12]

John J Hopfield. 1982. Neural networks and physical systems with emergent collective computational abilities.. In PNAS, Vol. 79. National Acad Sciences, 2554–2558. Proc. ACM Manag. Data, Vol. 3, No. 3 (SIGMOD), Article 209. Publication date: June 2025. RLOMM: An Efficient and ...

1982
[13]

Gang Hu, Jie Shao, Fenglin Liu, Yuan Wang, and Heng Tao Shen. 2017. IF-Matching: Towards Accurate Map-Matching with Information Fusion. In TKDE, Vol. 29. 114–127

2017
[14]

Hanwen Hu, Shiyou Qian, Jingchao Ouyang, Jian Cao, Han Han, Jie Wang, and Yirong Chen. 2023. AMM: An Adaptive Online Map Matching Algorithm. In TITS, Vol. 24. 5039–5051

2023
[15]

Zhenfeng Huang, Shaojie Qiao, Nan Han, Chang-an Yuan, Xuejiang Song, and Yueqiang Xiao. 2021. Survey on vehicle map matching techniques. In CAAI Trans. Intell. Technol., Vol. 6. 55–71

2021
[16]

Peter J. Huber. 1964. Robust Estimation of a Location Parameter. In The Annals of Mathematical Statistics , Vol. 35. 73–101

1964
[17]

George R Jagadeesh and Thambipillai Srikanthan. 2017. Online map-matching of noisy and sparse location data with hidden Markov and route choice models. In TITS, Vol. 18. IEEE, 2423–2434

2017
[18]

Linli Jiang, Chao-Xiong Chen, and Chao Chen. 2023. L2mm: learning to map matching with deep models for low-quality gps trajectory data. In TKDD, Vol. 17. ACM, 1–25

2023
[19]

Nan Jiang, Haitao Yuan, Jianing Si, Minxiao Chen, and Shangguang Wang. 2024. Towards Effective Next POI Prediction: Spatial and Semantic Augmentation with Remote Sensing Data. In ICDE. IEEE, 5061–5074

2024
[20]

Kingma and Jimmy Ba

Diederik P. Kingma and Jimmy Ba. 2015. Adam: A Method for Stochastic Optimization. In ICLR, Yoshua Bengio and Yann LeCun (Eds.)

2015
[21]

Kipf and Max Welling

Thomas N. Kipf and Max Welling. 2017. Semi-Supervised Classification with Graph Convolutional Networks. In ICLR

2017
[22]

Matej Kubicka, Arben Çela, Hugues Mounier, and Silviu-Iulian Niculescu. 2018. Comparative Study and Application- Oriented Classification of Vehicular Map-Matching Methods. In ITSM, Vol. 10. 150–166

2018
[23]

Biwei Liang, Tengjiao Wang, Shun Li, Wei Chen, Hongyan Li, and Kai Lei. 2016. Online Learning for Accurate Real-Time Map Matching. In PAKDD, James Bailey, Latifur Khan, Takashi Washio, Gillian Dobbie, Joshua Zhexue Huang, and Ruili Wang (Eds.), Vol. 9652. Springer, 67–78

2016
[24]

Zachary Chase Lipton. 2015. A Critical Review of Recurrent Neural Networks for Sequence Learning. In CoRR, Vol. abs/1506.00019

2015 arXiv
[25]

Yu Liu, Qian Ge, Wei Luo, Qiang Huang, Lei Zou, Haixu Wang, Xin Li, and Chang Liu. 2024. GraphMM: Graph-Based Vehicular Map Matching by Leveraging Trajectory and Road Correlations. In TKDE, Vol. 36. 184–198

2024
[26]

Yin Lou, Chengyang Zhang, Yu Zheng, Xing Xie, Wei Wang, and Yan Huang. 2009. Map-matching for low-sampling- rate GPS trajectories. In SIGSPATIAL, Divyakant Agrawal, Walid G. Aref, Chang-Tien Lu, Mohamed F. Mokbel, Peter Scheuermann, Cyrus Shahabi, and Ouri Wolfson (Eds.). ACM, 352–361

2009
[27]

Huali Lu, Feng Lyu, Huaqing Wu, Jie Zhang, Ju Ren, Yaoxue Zhang, and Xuemin Shen. 2023. FL-AMM: Federated Learning Augmented Map Matching With Heterogeneous Cellular Moving Trajectories. In IJSAC, Vol. 41. 3878–3892

2023
[28]

Riedmiller

Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Alex Graves, Ioannis Antonoglou, Daan Wierstra, and Martin A. Riedmiller. 2013. Playing Atari with Deep Reinforcement Learning. In CoRR, Vol. abs/1312.5602

2013 arXiv
[29]

Reham Mohamed, Heba Aly, and Moustafa Youssef. 2017. Accurate Real-time Map Matching for Challenging Environ- ments. In TITS, Vol. 18. 847–857

2017
[30]

Paul Newson and John Krumm. 2009. Hidden Markov map matching through noise and sparseness. In SIGSPATIAL, Divyakant Agrawal, Walid G. Aref, Chang-Tien Lu, Mohamed F. Mokbel, Peter Scheuermann, Cyrus Shahabi, and Ouri Wolfson (Eds.). ACM, 336–343

2009
[31]

Simon Aagaard Pedersen, Bin Yang, and Christian S. Jensen. 2020. Anytime Stochastic Routing with Hybrid Learning. In VLDB, Vol. 13. 1555–1567

2020
[32]

Lin Qu, Yue Zhou, Jiangxin Li, Qiong Yu, and Xinguo Jiang. 2023. HMM-Based Map Matching and Spatiotemporal Analysis for Matching Errors with Taxi Trajectories. In IJGI, Vol. 12. 330

2023
[33]

Mohammed A Quddus, Washington Y Ochieng, and Robert B Noland. 2007. Current map-matching algorithms for transport applications: State-of-the art and future research directions. In Transportation research part c: Emerging technologies, Vol. 15. Elsevier, 312–328

2007
[34]

Huimin Ren, Sijie Ruan, Yanhua Li, Jie Bao, Chuishi Meng, Ruiyuan Li, and Yu Zheng. 2021. MTrajRec: Map-Constrained Trajectory Recovery via Seq2Seq Multi-task Learning. In SIGKDD, Feida Zhu, Beng Chin Ooi, and Chunyan Miao (Eds.). ACM, 1410–1419

2021
[35]

Siavash Saki and Tobias Hagen. 2022. A Practical Guide to an Open-Source Map-Matching Approach for Big GPS Data. In SN Comput. Sci., Vol. 3. 415

2022
[36]

Zhihao Shen, Wan Du, Xi Zhao, and Jianhua Zou. 2020. DMM: fast map matching for cellular data. In MobiCom. ACM, 60:1–60:14

2020
[37]

Zhihao Shen, Kang Yang, Xi Zhao, Jianhua Zou, Wan Du, and Junjie Wu. 2024. DMM: A Deep Reinforcement Learning based Map Matching Framework for Cellular Data. In TKDE. IEEE

2024
[38]

Weijie Shi, Jiajie Xu, Junhua Fang, Pingfu Chao, An Liu, and Xiaofang Zhou. 2023. LHMM: A Learning Enhanced HMM Model for Cellular Trajectory Map Matching. In ICDE. IEEE, 2429–2442. Proc. ACM Manag. Data, Vol. 3, No. 3 (SIGMOD), Article 209. Publication date: June 2025. 209:26...

2023
[39]

Guy, Ming C

Avneesh Sud, Russell Gayle, Erik Andersen, Stephen J. Guy, Ming C. Lin, and Dinesh Manocha. 2008. Real-time navigation of independent agents using adaptive roadmaps. In SIGGRAPH. ACM, 56:1–56:10

2008
[40]

Shun Taguchi, Satoshi Koide, and Takayoshi Yoshimura. 2019. Online Map Matching With Route Prediction. In TITS, Vol. 20. 338–347

2019
[41]

Aäron van den Oord, Yazhe Li, and Oriol Vinyals. 2018. Representation Learning with Contrastive Predictive Coding. In CoRR, Vol. abs/1807.03748

2018 arXiv
[42]

Hado van Hasselt, Arthur Guez, and David Silver. 2016. Deep Reinforcement Learning with Double Q-Learning. In AAAI, Dale Schuurmans and Michael P. Wellman (Eds.). AAAI Press, 2094–2100

2016
[43]

Gomez, Lukasz Kaiser, and Illia Polosukhin

Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is All you Need. In NIPS. 5998–6008

2017
[44]

Jingyuan Wang, Ning Wu, Xinxi Lu, Wayne Xin Zhao, and Kai Feng. 2021. Deep Trajectory Recovery with Fine-Grained Calibration using Kalman Filter. In TKDE, Vol. 33. 921–934

2021
[45]

Adrian Wöltche. 2023. Open source map matching with Markov decision processes: A new method and a detailed benchmark with existing approaches. In Trans. GIS, Vol. 27. 1959–1991

2023
[46]

Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. 2019. How Powerful are Graph Neural Networks?. In ICLR

2019
[47]

Can Yang and Gyözö Gidófalvi. 2018. Fast map matching, an algorithm integrating hidden Markov model with precomputation. In IJGIS, Vol. 32. 547–570

2018
[48]

Haitao Yuan and Guoliang Li. 2021. A Survey of Traffic Prediction: from Spatio-Temporal Data to Intelligent Trans- portation. In Data Sci. Eng., Vol. 6. 63–85

2021
[49]

Haitao Yuan, Guoliang Li, and Zhifeng Bao. 2022. Route Travel Time Estimation on A Road Network Revisited: Heterogeneity, Proximity, Periodicity and Dynamicity. InVLDB, Vol. 16. 393–405

2022
[50]

Haitao Yuan, Guoliang Li, Zhifeng Bao, and Ling Feng. 2020. Effective Travel Time Estimation: When Historical Trajectories over Road Networks Matter. In SIGMOD, David Maier, Rachel Pottinger, AnHai Doan, Wang-Chiew Tan, Abdussalam Alawini, and Hung Q. Ngo (Eds.). ACM, 2135–2149

2020
[51]

Jing Yuan, Yu Zheng, Chengyang Zhang, Xing Xie, and Guangzhong Sun. 2010. An Interactive-Voting Based Map Matching Algorithm. In MDM, Takahiro Hara, Christian S. Jensen, Vijay Kumar, Sanjay Madria, and Demetrios Zeinalipour-Yazti (Eds.). IEEE Computer Society, 43–52

2010
[52]

Kai Zhao, Jie Feng, Zhao Xu, Tong Xia, Lin Chen, Funing Sun, Diansheng Guo, Depeng Jin, and Yong Li. 2019. DeepMM: Deep Learning Based Map Matching with Data Augmentation. In SIGSPATIAL, Farnoush Banaei Kashani, Goce Trajcevski, Ralf Hartmut Güting, Lars Kulik, and Shawn D. Ne...

2019
[53]

Kai Zheng, Yu Zheng, Xing Xie, and Xiaofang Zhou. 2012. Reducing Uncertainty of Low-Sampling-Rate Trajectories. In ICDE, Anastasios Kementsietsidis and Marcos Antonio Vaz Salles (Eds.). 1144–1155

2012
[54]

Zhihan Zheng, Haitao Yuan, Minxiao Chen, and Shangguang Wang. 2025. RLER-TTE: An Efficient and Effective Framework for En Route Travel Time Estimation with Reinforcement Learning. In SIGMOD, Vol. 3. 71:1–71:26

2025
[55]

Lei Zhu, Jacob R Holden, and Jeffrey D Gonder. 2017. Trajectory segmentation map-matching approach for large-scale, high-resolution GPS data. In Transportation Research Record, Vol. 2645. 67–75. Received October 2024; revised January 2025; accepted February 2025 Proc. ACM Mana...

2017

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.

[1] [1]

2015. kaggle. https://www.kaggle.com/c/pkdd-15-taxi-trip-time-prediction-ii

work page 2015

[2] [2]

2021. GAIA. https://outreach.didichuxing.com/research/opendata/

work page 2021

[3] [3]

Pingfu Chao, Yehong Xu, Wen Hua, and Xiaofang Zhou. 2020. A Survey on Map-Matching Algorithms. InADC, Renata Borovica-Gajic, Jianzhong Qi, and Weiqing Wang (Eds.), Vol. 12008. Springer, 121–133

work page 2020

[4] [4]

Minxiao Chen, Haitao Yuan, Nan Jiang, Zhifeng Bao, and Shangguang Wang. 2024. Urban Traffic Accident Risk Prediction Revisited: Regionality, Proximity, Similarity and Sparsity. InCIKM. ACM, 281–290

work page 2024

[5] [5]

Kyunghyun Cho, Bart van Merrienboer, Çaglar Gülçehre, Dzmitry Bahdanau, Fethi Bougares, Holger Schwenk, and Yoshua Bengio. 2014. Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation. In EMNLP, Alessandro Moschitti, Bo Pang, and Walter Daelemans (Eds.). ACL, 1724–1734

work page 2014

[6] [6]

Sean R Eddy. 1996. Hidden markov models. In Current opinion in structural biology , Vol. 6. Elsevier, 361–365

work page 1996

[7] [7]

Jie Feng, Yong Li, Kai Zhao, Zhao Xu, Tong Xia, Jinglin Zhang, and Depeng Jin. 2022. DeepMM: Deep Learning Based Map Matching With Data Augmentation. In TMC, Vol. 21. 2372–2384

work page 2022

[8] [8]

Xiao Fu, Jiaxu Zhang, and Yue Zhang. 2021. An Online Map Matching Algorithm Based on Second-Order Hidden Markov Model. In JAT, Vol. 2021. 9993860

work page 2021