RuDaCoP: The Dataset for Smartphone-based Intellectual Pedestrian Navigation

Alexey Derevyankin; Alexey Nikulin; Andrey Bayev; Ilya Gartseev; Ivan Chistyakov; Mikhail Pikhletsky

REVIEW 4 major objections 5 minor 19 references

RuDaCoP: The Dataset for Smartphone-based Intellectual Pedestrian Navigation

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read 1,200 smartphone walks now come with foot-level ground truth

desk verdict A large and carefully documented smartphone inertial dataset, but the 'high accuracy ground truth' claim is only internally validated and needs external support before the dataset can serve as absolute reference. read the letter →

arxiv 1908.03609 v1 pith:AKSETIFE submitted 2019-08-09 eess.SY cs.LGcs.SY

Andrey Bayev , Ilya Gartseev , Ivan Chistyakov , Alexey Nikulin , Alexey Derevyankin , Mikhail Pikhletsky This is my paper · ORCID

classification eess.SYcs.LGcs.SY

keywords pedestriandeadreckoningsmartphoneinertialsensorsdatasetgroundtruthfoot-mountedIMUZUPT-aidednavigationmachinelearningindoor

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 introduces RuDaCoP, a dataset of about 1,200 recordings of smartphone inertial sensors, collected while more than a hundred people walked flat, closed-loop trajectories of one to ten minutes. Each recording is accompanied by reference data, including per-foot trajectories, a fused center-of-gravity trajectory, stationary indicators, and per-step lengths and headings, computed from two foot-mounted IMUs. The authors argue that this fills a gap: existing inertial datasets either lack trajectory-level ground truth at the accuracy needed for step-length estimation or cover too little variety in people, devices, and smartphone placements. If the reference data are reliable, the dataset would give both classical pedestrian dead-reckoning developers and machine-learning practitioners a common benchmark and training resource.

What carries the argument

The core object is the dataset itself, organized as core data, raw accelerometer, gyroscope, and magnetometer readings from several smartphone models in calibrated and uncalibrated forms, plus reference data, left and right foot trajectories, stationary flags, step lengths and headings, and a fused center-of-gravity trajectory sampled at 125 Hz. The load-bearing mechanism that generates the reference is ZUPT-aided inertial navigation: a Kalman-filter reconstruction enforces zero velocity whenever the stance detector decides the foot is motionless, and enforces the closed-loop condition that the final position equals the initial position; Rauch-Tung-Striebel smoothing then removes discontinuities in the reconstructed trajectory. Dynamic time warping is used to align the heading frames of the two feet and to verify that the independently reconstructed foot trajectories are consistent.

What would settle it

Run a subset of the closed-loop walks under an optical motion-capture system that tracks both feet at centimeter-level accuracy and compare its trajectories to the RuDaCoP reference files; if the differences are much larger than the claimed foot-position tolerance or exceed typical step-length errors, the reference data would not be high-accuracy. A simpler check is to compute, for every experiment, the reconstructed final foot position against the paper-marked start position and see whether the closure error is actually within five centimeters for each foot.

Watch

Extended reading notes

Core claim

The central claim is that RuDaCoP supplies high-accuracy, trajectory-level ground truth for smartphone-based pedestrian navigation at a scale and diversity that earlier datasets lack. The ground truth is produced by post-processing two foot-mounted IMUs with a ZUPT-aided Kalman filter, Rauch-Tung-Striebel smoothing, and a joint left-right foot fusion that uses the prior knowledge that every trajectory is flat and closed-loop. The paper also details the supporting algorithms: a stance detector based on zero-velocity detection, a step detector that marks the start of each motion, and a DTW-based consistency check that compares the two foot trajectories and rejects suspicious experiments. Two usage examples demonstrate the dataset: a smartphone step-duration detector whose output tracks the reference, and a convolutional neural network trained on raw phone accelerometer and gyroscope data to predict walking speed, with an average distance error of about 3 percent.

Load-bearing premise

The reference trajectories are only as accurate as the assumptions that every foot-on-ground period is correctly detected as motionless, that the floor is truly flat, that the walk genuinely returns to its start within about five centimeters per foot, and that the foot-mounted IMUs remain rigidly attached throughout the walk.

Editorial extensions

If this is right

Researchers can train a step detector or step-length estimator on phone IMU data using the per-foot stationary markers and step files as supervision, without building their own motion-capture rig.
Classical PDR algorithms can be benchmarked against the fused center-of-gravity trajectory on a common set of walks spanning ages 18 to 60, multiple smartphone models, and placements from pockets to bags to hands.
Because the ground-truth generation pipeline is specified in detail, the reference files can be regenerated or extended for new walks collected under the same flat closed-loop protocol.
The reported 3 percent distance error for a CNN speed predictor indicates that the dataset is usable for learning-based speed and step-length estimation, not only for evaluating classical approaches.

Reading between the lines

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

A natural and unstated next step is to define an official train/test split that separates participants, paths, and phone models, since the paper does not describe one and the value for machine learning will depend on avoiding leakage between training and evaluation.
The flat closed-loop restriction means the dataset cannot directly validate algorithms for stairs, slopes, or unrestricted indoor routes; adding a companion set with three-dimensional trajectories and optical ground truth would extend its reach.
The DTW-based consistency metric could be published per experiment as a quality score, letting users down-weight borderline trajectories rather than discarding them entirely.
The same collection protocol could be adapted to carry a foot-mounted camera or visual odometry unit alongside the IMUs, which would let ground truth be produced in arbitrary three-dimensional environments rather than only flat closed loops.

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

Two things to know about this paper. First, the dataset itself is a genuinely useful contribution: about 1,200 smartphone inertial recordings from four phones per experiment, over 100 participants, varied phone placements, shoes, and floor types, with dual foot-mounted IMU reference trajectories. That combination is not present in the prior datasets I know of. Second, the 'high accuracy ground truth' claim is not actually demonstrated. The only quality check is the DTW distance between left and right foot trajectories, which is an internal consistency measure and cannot catch common-mode errors like yaw drift or gyro bias. There is no external reference (motion capture, surveyed path, etc.) reported anywhere.

The collection pipeline is described well: hardware, mounting, Bluetooth synchronization, file formats, and metadata are all concrete enough for a researcher to use the data without much guesswork. The reference trajectory processing uses standard ZUPT-aided Kalman smoothing with left/right fusion, which is not new but is applied at a scale and diversity that makes the dataset valuable for training and benchmarking.

The soft spots, in order of importance. First, the accuracy of the ground truth is the load-bearing claim, and it is only internally validated. The algorithm enforces a closed-loop constraint with p_final = 0 even though the paper admits start and finish positions differ by up to 5 cm per foot. That means the endpoint is forced to zero, which can warp the interior of longer trajectories. The DTW metric between feet cannot detect a bias that affects both feet equally. Without an independent accuracy evaluation, the dataset's value as absolute ground truth is unproven. That's not a reason to reject the paper, but it is a reason to require a fix. Second, the speed prediction example reports a 3% error but gives almost no detail about the network architecture, training/test split, or amount of data used, so that example is not independently evaluable.

The paper is honest about the flatness and closure assumptions, and the related work survey is reasonable. The dataset appears to be a real resource for the community, and the flaws are addressable: an external validation on a subset of experiments, or at least a sensitivity analysis of the ZUPT detector thresholds and the closure constraint, would change my confidence a lot.

For anyone working on smartphone PDR or inertial navigation, this is worth a careful look. I'd bring it to a reading group and cite it if I was working on learning-based pedestrian navigation. The paper should go to peer review, but I would not accept it in its current form; the ground truth accuracy claim needs to be backed by evidence beyond internal consistency.

Referee Report

4 major / 5 minor

Summary. The paper presents RuDaCoP, a dataset of approximately 1200 smartphone inertial recordings (accelerometer, gyroscope, magnetometer) collected from over one hundred participants during 1–10 minute closed-loop walks on flat surfaces, with a variety of smartphone placements, footwear, flooring, and walking speeds. The reference data consist of left-foot, right-foot, and center-of-gravity trajectories at 125 Hz, plus step and stance labels, generated by two foot-mounted IMUs processed by a ZUPT-aided dual-foot Kalman filter with RTS smoothing and loop-closure constraints. The paper describes the hardware, collection protocol, synchronization, file formats, reconstruction algorithm, and two application examples (step detection and neural-network speed prediction). The central claim is that the reference data are 'high accuracy ground truth' suitable for training and benchmarking smartphone PDR algorithms.

Significance. Should the dataset be reliable, it would be a valuable community resource: it is considerably larger and more diverse than most existing public datasets for smartphone pedestrian navigation, includes raw and calibrated sensor data from multiple phone models, and provides step- and stance-level labels alongside trajectories. The paper's detailed description of the collection pipeline, synchronization scheme, and file formats is a strength and should enable others to use the data. However, the significance is currently limited by the lack of any independent, absolute accuracy assessment of the reference trajectories; the internal consistency measure reported is not sufficient to support the 'high accuracy' claim.

major comments (4)

[Section IV-A (and Abstract/Section I)] The 'high accuracy ground truth' claim is unsupported by an external, absolute error estimate. The only quantitative quality metric is the DTW distance between the left and right foot trajectories (Fig. 6), which is an internal consistency check and cannot detect common-mode errors such as yaw drift, gyro bias, or timing errors that affect both feet similarly. Moreover, this metric is not independent: the algorithm itself uses DTW minimization to determine the relative heading between the feet (Section IV-A), so the reported DTW values are biased toward small values. The authors should either validate the reference trajectories against an independent surveyed reference (e.g., known path lengths, total station, or motion capture) for a subset of experiments and report the error, or explicitly rephrase the claim to 'reference data with internally consistent reconstruction' and discuss the possible common-mode error bounds.
[Section III-A and IV-A] The reconstruction imposes the closed-loop constraint p_final = 0, but Section III-A states that the start and finish positions differ by up to 5 cm per foot. This constraint therefore introduces a known, unmodeled bias. For 10-minute trajectories, forcing the endpoint to zero can warp the interior of the reconstructed path, and the magnitude of this distortion is not assessed. The authors should quantify the effect (e.g., by comparing reconstructions with and without the constraint or by estimating the induced error) or justify why it is negligible for the claimed accuracy.
[Section IV-A] The trajectory reconstruction is based on the authors' own prior algorithm [16] and assumes flat horizontal motion, rigid foot mounting of the IMUs, reliable zero-velocity detection, and accurate time synchronization. No independent validation of the algorithm against an external reference is provided. Since the dataset is intended for training learning-based methods, any systematic bias in the reference labels (e.g., from incorrect zero-velocity detection) will be learned by downstream models. The paper should include a validation experiment or at least a careful sensitivity analysis of the reconstruction to failures of these assumptions.
[Section IV-B] The foot-motionless detection and step detection parameters (epsilon = 0.5 rad/s, alpha = 0.25, minStillTime, gamma, window h, and the covariance matrices Q, R', R'') are given with the note that they 'may be slightly varied,' but no sensitivity analysis is reported. Because stance and step labels are part of the reference data and directly influence the trajectory reconstruction, the authors should report the sensitivity of the reconstructed trajectories and labels to these parameters, or justify that the chosen values are robust.

minor comments (5)

[Abstract] The abstract contains several spacing errors, e.g., 'smartphone s' should be 'smartphones'; the text should be proofread for similar typographical issues.
[Figure 6] The plot lacks clear axis labels and units; please specify that the horizontal axis is DTW distance (in meters, presumably) and the vertical axis is the number of experiments.
[Section III-C] The claim that the time offset between smartphones is 'under 1 ms for most cases' should be supported by a measurement or at least a brief explanation of how this was empirically estimated.
[Section V] The step-duration comparison in Fig. 8 is only qualitative; a quantitative error metric (e.g., mean absolute error) would clarify the agreement between smartphone-based and reference step durations.
[Section I] The term 'intellectual pedestrian navigation' is nonstandard and potentially confusing; consider replacing it with 'learning-based pedestrian navigation' or define it on first use.

Circularity Check

1 steps flagged · score 4.0 of 10

Reference-data accuracy rests on the authors' own processing chain and an internal left/right DTW check, so the 'high accuracy ground truth' claim is not independently established.

self citation load bearing [Section IV-A (Reconstruction of trajectories) and reference [16]]
"In order to reconstruct the presented trajectories, we used the algorithms of a Kalman filter type. The algorithms are described in details in [16]. ... After that, we use the algorithm for construction of two "fused" trajectories as described in [16] ... We estimate the quality of algorithms by proximity of feet trajectories that were obtained in same experiments."

The reference trajectories—the 'high accuracy ground truth' promised in the Abstract—are produced by the authors' own Kalman-filter/ZUPT pipeline, with details cited to the same authors' prior paper [16] (Chistiakov, Nikulin, Gartseev). The only in-paper validation is the DTW proximity between left- and right-foot trajectories, both outputs of the same algorithm and subject to the same flat-loop and zero-final-position assumptions. This is internal consistency, not external calibration: common-mode errors (yaw drift, bias, timing, stance-detection errors) affect both feet and would not raise the DTW distance. Thus the accuracy claim for the labels reduces to the correctness of the authors' own algorithm rather than to any independent measurement.

full rationale

The core dataset (roughly 1200 raw smartphone IMU recordings) is an independent artifact: those measurements are not derived from the reference labels, and the paper is transparent that the reference trajectories are post-processed by the authors' algorithms (Sections I and IV-A). No equation-level derivation in the paper equates a predicted quantity to a fitted input. However, the dataset's central value as a benchmark depends on the reference trajectories being accurate in an absolute sense, and that accuracy is supported mainly by (i) a self-citation to the same authors' prior work [16] for the reconstruction algorithm and (ii) an internal left/right DTW consistency metric (Fig. 6) plus loop-closure enforcement. Neither of these can certify absolute trajectory accuracy; they are self-consistency checks inside the production chain. This makes the 'high accuracy ground truth' claim partially self-referential, though not a formal circular derivation. The neural-network speed-prediction example in Section V is too under-specified (no train/test split is stated) to warrant a separate fitted-input-called-prediction charge without speculation.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The central dataset does not introduce new physical entities. It relies on the domain assumptions of flat closed-loop trajectories, rigid foot mounting, reliable stance detection, and time synchronization, plus hand-set detector parameters and unspecified filter covariances that all affect the accuracy of the reference data.

free parameters (5)

epsilon = 0.5 rad/s
Hand-set threshold in the foot stationary detection rule in Section IV-B; directly controls which samples are labeled as stationary.
alpha = 0.25
Hand-set verticality threshold in the foot stationary detection rule in Section IV-B; affects ground-truth stationary labels.
minStillTime = not stated
Step detection parameter in Figure 7; the text says it may be varied but does not give its value for the released dataset.
gamma and window h in ZUPT detector = not stated
The zero-velocity detector from [17] is used with a fixed constant gamma and window size h, but neither value is reported in the paper.
Q, R', R'' covariance matrices = not stated
Covariance matrices for the Kalman and RTS filters in the reconstruction algorithm shown in Figure 4 are required but their values are not given.

assumptions (6)

domain assumption All trajectories are closed loops with start and finish positions within 5 cm, enforced by paper markers.
Section III-A defines this restriction and Section IV-A uses it as the final position constraint p_final = 0 in the reconstruction algorithm.
domain assumption All trajectories are flat horizontal surfaces with no stairs or significant height changes.
Section III-A restricts trajectories to flat ground, which justifies the 2D horizontal representation of positions and steps.
domain assumption Foot-mounted IMU modules remain immobile relative to the foot during walking.
Section III-B states the modules are mounted with rubber bindings to keep them immobile; this is required for ZUPT updates to be valid.
domain assumption The zero-velocity detector correctly identifies all motionless phases of each foot.
Section IV-A and IV-B rely on the detector output S_n to enforce zero-velocity constraints; errors propagate directly into the reference trajectories.
domain assumption Android timestamps and Bluetooth delays allow alignment of all devices to a common time grid with error under 1 ms.
Section III-C states the synchronization procedure and the empirical estimate; core and reference data are aligned using these timestamps.
standard math Kalman filtering, RTS smoothing, and dynamic time warping are valid for the reconstructed linearized system.
Section IV-A applies standard estimation methods to the linearized inertial navigation system without proving their convergence or error bounds.

how reviews work

0 comments

Cite this review

Pith. "Pith review of RuDaCoP: The Dataset for Smartphone-based Intellectual Pedestrian Navigation." pith.science (2026). https://pith.science/paper/AKSETIFE

@misc{pith2026190803609,
  author       = {Pith},
  title        = {Pith review of: RuDaCoP: The Dataset for Smartphone-based Intellectual Pedestrian Navigation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AKSETIFE}},
  note         = {Machine review of arXiv:1908.03609}
}

read the original abstract

This paper presents the large and diverse dataset for development of smartphone-based pedestrian navigation algorithms. This dataset consists of about 1200 sets of inertial measurements from sensors of several smartphones. The measurements are collected while walking through different trajectories up to 10 minutes long. The data are accompanied by the high accuracy ground truth collected with two foot-mounted inertial measurement units and post-processed by the presented algorithms. The dataset suits both for training of intellectual pedestrian navigation algorithms based on learning techniques and for development of pedestrian navigation algorithms based on classical approaches. The dataset is accessible at http://gartseev.ru/projects/ipin2019.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 18 canonical work pages

[16]

Pedestrian dead - reckoning algorithms for dual foot-mounted inertial sensors,

I. A. Chistiakov, A. A. Nikulin, I. B. Gartseev, “Pedestrian dead - reckoning algorithms for dual foot-mounted inertial sensors,” in Proc. 25th Saint P etersburg International Conference on Integrated Navigation Systems (ICINS), May 2019, pp. 1–8

work page 2019
[1]

An introduction to inertial navigation,

O. J. Woodman, “An introduction to inertial navigation,” University of Cambridge Computer Laboratory, Tech. Rep. 696, 2007

work page 2007
[2]

Improved pedestrian dead reckoning positioning with gait parameter learning,

P. Kasebzadeh, C. Fritsche, G. Hendeby, F. Gunnarsson, and F. Gustafsson, “Improved pedestrian dead reckoning positioning with gait parameter learning,” in International Conference on Information Fusion, Heidelberg, Germany, Jul. 2016

work page 2016
[3]

OxIOD: The Dat aset for Deep Inertial Odometry,

C. Chen, P. Zhao, C. X. Lu et al., “OxIOD: The Dat aset for Deep Inertial Odometry,” in arXiv 1809.07491, 2018

arXiv 2018
[4]

Ionet: Learning to cure the curse of drift in inertial odometry,

C. Chen, X. Lu, A. Markham, and N. Trigoni, “Ionet: Learning to cure the curse of drift in inertial odometry,” in Proceedings of the Thirty - Second AAAI Conference on Artificial Intelligence, New Orleans, Louisiana, USA, February 2–7, 2018

work page 2018
[5]

Step Length Estimation Methods Based on Inertial Sensors: A Review,

L. E. Díez, A. Bahillo, J. Otegui, and T. Otim, “Step Length Estimation Methods Based on Inertial Sensors: A Review,” IEEE Sensors Journal, vol. 18, no. 17, pp. 6908–6926, September 2018

work page 2018
[6]

IMU Dataset For Motion and Device Mode Classification,

P. Kasebzadeh, G. Hendeby, C. Fritsche, F. Gunnarsson, and F. Gustafsson, “ IMU Dataset For Motion and Device Mode Classification,” in Proc. Int. Conf. Indoor Positioning Indoor Navigat. (IPIN), Sep. 2017, pp. 1–8

work page 2017
[7]

HASC-IPSC: Indoor Pedestrian Sensing Corpus with a Balance of Gender and Age for Indoor Positioning and Floor-plan Generation Researches,

K. Kaji, H. Watanabe, R. Ban, and N. Kawaguchi, “ HASC-IPSC: Indoor Pedestrian Sensing Corpus with a Balance of Gender and Age for Indoor Positioning and Floor-plan Generation Researches,” in Proc. ACM Conference on Pervasive and Ubiquitous Computing Adjunct Publication, Sep. 2013, pp.605–610

work page 2013

Show all 19 references

[8]

Detecting Falls with Wearable Sensors Using Machine Learning Techniques ,

A.T. Özdemir, B. Barsh an, “Detecting Falls with Wearable Sensors Using Machine Learning Techniques ,” Sensors, vol. 14, pp. 10691– 10708, 2014

2014
[9]

WeAllWalk: An Annotated Data Set of Inertial Sensor Time Series from Blind Walkers ,

G. H. Flores , R. Manduchi, “WeAllWalk: An Annotated Data Set of Inertial Sensor Time Series from Blind Walkers ,” in Proc. 18th International ACM SIGACCESS Conference on Computers and Accessibility, Oct. 2016, pp. 141–150

2016
[10]

Walk detection and step counting on unconstrained smartphones ,

A. Brajdic, R. Harle., “Walk detection and step counting on unconstrained smartphones ,” in Proc. ACM international joint conference on Pervasive and ubiquitous computing , 2013, pp. 225– 234

2013
[11]

The largest inertial sensor -based gait database and performance evaluation of gait-based personal authentication,

T. T. Ngo, Y. Makihara, H. Nagahara, Y. Mukaigawa, and Y . Yagi, “The largest inertial sensor -based gait database and performance evaluation of gait-based personal authentication,” Pattern Recognition 47, 1 (2014), pp. 228–237

2014
[12]

Interiornet: Mega-scale multi- sensor photo-realistic indoor scenes dataset,

W. Li, S. Saeedi, J. McCormac, R. Clark, D. Tzoumanikas, Q. Ye, Y. Huang, R. Tang, and S. Leutenegger. “Interiornet: Mega-scale multi- sensor photo-realistic indoor scenes dataset,” in British Machine Vis. Conf. (BMVC), Sept. 2018

2018
[13]

Foot -mounted INS for everybody – an open-source embedded implementation,

J.-O. Nilsson, I. Skog, P. Handel, and K. V. S. Hari, “Foot -mounted INS for everybody – an open-source embedded implementation,” in Proc. of the 2012 IEEE/ION Position, Location and Navigation Symposium, pp. 140–145

2012
[14]

OSMIUM MIMU22BL/22BLP/22BLPX, www.inertialelements.com/ documents/mimu22bl/MIMU22BL_product-brief.pdf
[15]

MPU-9250 Product Specification, http://www.invensense.com/wp- content/uploads/2015/02/PS-MPU-9250A-01-v1.1.pdf

2015
[17]

Zero -velocity detection: An algorithm evaluation,

I. Skog, P. Handel, J.- O. Nilsson, and J. Rantakokko, “Zero -velocity detection: An algorithm evaluation,” Biomedical Engineering, IEEE Transactions on, vol. 57, pp. 2657–2666, nov. 2010

2010
[18]

Smoothing for ZUPT-aided INSs,

D. S. Colomar, J.-O. Nilsson, P. Handel, and K. V. S. Hari, “Smoothing for ZUPT-aided INSs,” in Proc. International Conference on Indoor Positioning and Indoor Navigation (IPIN), 2012, pp. 1–5

2012
[19]

Dynamic Time Warping ,

M. Müller, “Dynamic Time Warping ,” in Information Retrieval for Music and Motion, Springer, 2007, pp. 69–84. Fig. 9. Sample output of pedestrian speed prediction neural network (upper four graphs), histogram of error ratio (lowermost)

2007

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [16]

Pedestrian dead - reckoning algorithms for dual foot-mounted inertial sensors,

I. A. Chistiakov, A. A. Nikulin, I. B. Gartseev, “Pedestrian dead - reckoning algorithms for dual foot-mounted inertial sensors,” in Proc. 25th Saint P etersburg International Conference on Integrated Navigation Systems (ICINS), May 2019, pp. 1–8

work page 2019

[2] [1]

An introduction to inertial navigation,

O. J. Woodman, “An introduction to inertial navigation,” University of Cambridge Computer Laboratory, Tech. Rep. 696, 2007

work page 2007

[3] [2]

Improved pedestrian dead reckoning positioning with gait parameter learning,

P. Kasebzadeh, C. Fritsche, G. Hendeby, F. Gunnarsson, and F. Gustafsson, “Improved pedestrian dead reckoning positioning with gait parameter learning,” in International Conference on Information Fusion, Heidelberg, Germany, Jul. 2016

work page 2016

[4] [3]

OxIOD: The Dat aset for Deep Inertial Odometry,

C. Chen, P. Zhao, C. X. Lu et al., “OxIOD: The Dat aset for Deep Inertial Odometry,” in arXiv 1809.07491, 2018

arXiv 2018

[5] [4]

Ionet: Learning to cure the curse of drift in inertial odometry,

C. Chen, X. Lu, A. Markham, and N. Trigoni, “Ionet: Learning to cure the curse of drift in inertial odometry,” in Proceedings of the Thirty - Second AAAI Conference on Artificial Intelligence, New Orleans, Louisiana, USA, February 2–7, 2018

work page 2018

[6] [5]

Step Length Estimation Methods Based on Inertial Sensors: A Review,

L. E. Díez, A. Bahillo, J. Otegui, and T. Otim, “Step Length Estimation Methods Based on Inertial Sensors: A Review,” IEEE Sensors Journal, vol. 18, no. 17, pp. 6908–6926, September 2018

work page 2018

[7] [6]

IMU Dataset For Motion and Device Mode Classification,

P. Kasebzadeh, G. Hendeby, C. Fritsche, F. Gunnarsson, and F. Gustafsson, “ IMU Dataset For Motion and Device Mode Classification,” in Proc. Int. Conf. Indoor Positioning Indoor Navigat. (IPIN), Sep. 2017, pp. 1–8

work page 2017

[8] [7]

HASC-IPSC: Indoor Pedestrian Sensing Corpus with a Balance of Gender and Age for Indoor Positioning and Floor-plan Generation Researches,

K. Kaji, H. Watanabe, R. Ban, and N. Kawaguchi, “ HASC-IPSC: Indoor Pedestrian Sensing Corpus with a Balance of Gender and Age for Indoor Positioning and Floor-plan Generation Researches,” in Proc. ACM Conference on Pervasive and Ubiquitous Computing Adjunct Publication, Sep. 2013, pp.605–610

work page 2013