pith. sign in

arxiv: 1907.09454 · v1 · pith:LB2YUPOMnew · submitted 2019-07-16 · 📡 eess.SP · cs.NI

A Fog Computing Framework for Autonomous Driving Assist: Architecture, Experiments, and Challenges

Muthucumaru Maheswaran , Tianzi Yang , Salman Memon This is my paper

Pith reviewed 2026-05-24 20:29 UTC · model grok-4.3

classification 📡 eess.SP cs.NI
keywords fog computingautonomous drivingdigital twinsmachine learning forecastinghazard mappingedge computingvehicle trackingmaneuver suggestion
0
0 comments X

The pith

A fog computing framework builds edge digital twins from camera and sensor data, uses machine learning to forecast vehicle positions, and applies a box algorithm to map hazards for safe autonomous maneuvers.

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

The paper proposes a fog computing framework that creates a network of distributed digital twins called edge twins on fog machines, updated continuously with vehicle locations from overhead cameras and sensor feeds. A machine learning forecaster predicts future positions to handle communication delays, while a box algorithm turns those forecasts into a hazard map that recommends maneuvers such as lane changes and accelerations. The approach is evaluated through simulations on real-world highway vehicle position traces, aiming to provide global road space allocation for mixed autonomous and human-driven traffic.

Core claim

The framework relies on overhead views from cameras and data streams from vehicle sensors to create a network of distributed digital twins on fog machines. The edge twin is continuously updated with the locations of both autonomous and human-piloted vehicles. A machine learning forecaster predicts future vehicle locations to address communication delays, and a box algorithm uses the forecasts to create a hazard map for suggesting safe maneuvers.

What carries the argument

The edge twin, a distributed digital twin on fog machines that harvests vehicle locations, applies machine learning forecasting for position predictions, and runs a box algorithm to produce hazard maps.

If this is right

  • The edge twin can allocate road space from a global viewpoint while the forecaster compensates for delays in data reaching the fog machines.
  • Hazard maps derived from forecasted locations enable concrete suggestions for lane changes and accelerations.
  • Simulations based on real-world vehicle position traces validate the location harvesting, forecasting, and mapping components.
  • The framework supports both autonomous and human-piloted vehicles on road segments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same edge-twin structure could be extended to coordinate multiple road segments in sequence.
  • Replacing the box algorithm with alternative mapping methods might change the precision of suggested maneuvers.
  • Adding direct vehicle-to-fog links could reduce the reliance on overhead cameras for initial location data.

Load-bearing premise

Vehicle locations harvested from overhead cameras and sensor feeds, combined with machine learning forecasts, can sufficiently overcome communication delays to enable reliable hazard mapping and safe maneuver suggestions.

What would settle it

A simulation run on the real highway dataset where the forecaster's predicted positions produce hazard maps that recommend maneuvers leading to simulated collisions.

Figures

Figures reproduced from arXiv: 1907.09454 by Muthucumaru Maheswaran, Salman Memon, Tianzi Yang.

Figure 1
Figure 1. Figure 1: Main task flows of auto-drive assist in the edge [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: Illustrating the importance of time shifting in the [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: A comparison of the predicted output from the tra [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: A comparison of the predicted output from the tra [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: A comparison of the predicted output from the tra [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: A box plot of the error (called the coordinate dis [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: A box plot of the error (called the coordinate dis [PITH_FULL_IMAGE:figures/full_fig_p007_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: This figure depicts the box algorithm output over [PITH_FULL_IMAGE:figures/full_fig_p008_9.png] view at source ↗
read the original abstract

Autonomous driving is expected to provide a range of far-reaching economic, environmental and safety benefits. In this study, we propose a fog computing based framework to assist autonomous driving. Our framework relies on overhead views from cameras and data streams from vehicle sensors to create a network of distributed digital twins, called an edge twin, on fog machines. The edge twin will be continuously updated with the locations of both autonomous and human-piloted vehicles on the road segments. The vehicle locations will be harvested from overhead cameras as well as location feeds from the vehicles themselves. Although the edge twin can make fair road space allocations from a global viewpoint, there is a communication cost (delay) in reaching it from the cameras and vehicular sensors. To address this, we introduce a machine learning forecaster as a part of the edge twin which is responsible for predicting the future location of vehicles. Lastly, we introduce a box algorithm that will use the forecasted values to create a hazard map for the road segment which would be used by the framework to suggest safe manoeuvres for the autonomous vehicles such as lane changes and accelerations. We present the complete fog computing framework for autonomous driving assist and evaluate key portions of the proposed framework using simulations based on a real-world dataset of vehicle position traces on a highway

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes a fog computing framework for autonomous driving assistance. It creates a network of distributed digital twins (edge twins) on fog machines, updated continuously with vehicle locations from overhead cameras and onboard sensors. A machine learning forecaster predicts future vehicle positions to offset communication delays, and a box algorithm generates hazard maps to recommend safe maneuvers such as lane changes and accelerations. Key portions of the framework are evaluated through simulations on a real-world highway vehicle position dataset.

Significance. If the simulation results demonstrate that the forecaster and hazard mapping reliably compensate for delays, the work could contribute to practical edge-assisted autonomous driving systems in mixed human/autonomous traffic. The use of real vehicle traces for evaluation is a strength that grounds the architecture in observable data.

major comments (2)
  1. [Evaluation / Experiments] The abstract and evaluation description state that key portions are evaluated via simulations on real traces, but no quantitative metrics (e.g., forecaster prediction error over specific time horizons, hazard map accuracy, or maneuver suggestion success rates) or baseline comparisons are reported. This is load-bearing for the central claim that the framework overcomes communication delays.
  2. [Framework description] The box algorithm is introduced to create hazard maps from forecasted locations, but the manuscript provides no algorithmic details, pseudocode, or parameter definitions (e.g., box size, hazard thresholds). This prevents assessment of how the hazard map translates into maneuver suggestions.
minor comments (2)
  1. [Introduction] Notation for 'edge twin' and 'box algorithm' is introduced without prior definition or reference to related work on digital twins in vehicular networks.
  2. [Architecture] The manuscript would benefit from a diagram or pseudocode clarifying the data flow between cameras, sensors, forecaster, and box algorithm.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which identify key areas where the manuscript can be strengthened. We agree that both the evaluation and the box algorithm require more detail to support the central claims. We address each major comment below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Evaluation / Experiments] The abstract and evaluation description state that key portions are evaluated via simulations on real traces, but no quantitative metrics (e.g., forecaster prediction error over specific time horizons, hazard map accuracy, or maneuver suggestion success rates) or baseline comparisons are reported. This is load-bearing for the central claim that the framework overcomes communication delays.

    Authors: We agree that the current evaluation lacks the specific quantitative metrics and baseline comparisons needed to fully substantiate the delay-compensation claims. In the revised manuscript we will add explicit results, including forecaster prediction error as a function of time horizon, hazard-map accuracy, maneuver success rates, and comparisons against baselines that do not use forecasting. revision: yes

  2. Referee: [Framework description] The box algorithm is introduced to create hazard maps from forecasted locations, but the manuscript provides no algorithmic details, pseudocode, or parameter definitions (e.g., box size, hazard thresholds). This prevents assessment of how the hazard map translates into maneuver suggestions.

    Authors: The box algorithm was presented at a high level. We will expand the description in the revision to include the full algorithmic procedure, pseudocode, definitions of all parameters (box size, hazard thresholds, etc.), and a clear mapping from hazard-map values to the suggested maneuvers. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper is an architecture proposal for a fog computing framework. It describes components (edge twin, ML forecaster, box algorithm) and evaluates them via simulation on an external real-world vehicle position dataset. No equations, derivations, fitted parameters, or self-citation chains appear that reduce any claimed result to its own inputs by construction. The central claims are design and empirical evaluation statements, not mathematical predictions forced by internal definitions or prior self-citations.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 2 invented entities

Based solely on the abstract; no free parameters, axioms, or independent evidence for new entities are specified.

invented entities (2)
  • edge twin no independent evidence
    purpose: network of distributed digital twins on fog machines updated with vehicle locations
    Introduced as a core component of the framework without external validation or falsifiable handle mentioned.
  • box algorithm no independent evidence
    purpose: uses forecasted values to create hazard map for suggesting maneuvers
    New component introduced to operationalize the forecasts, no independent evidence provided.

pith-pipeline@v0.9.0 · 5764 in / 1166 out tokens · 26258 ms · 2026-05-24T20:29:47.252934+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages · 5 internal anchors

  1. [1]

    Fog computing and its role in the internet of things,

    F. Bonomi, R. Milito, J. Zhu, and S. Addepalli, “Fog computing and its role in the internet of things, ” inProceedings of the first edition of the MCC workshop on Mobile cloud computing, pp. 13–16, ACM, 2012

    work page 2012

  2. [2]

    Shaping the digital twin for design and production engineering,

    B. Schleich, N. Anwer, L. Mathieu, and S. Wartzack, “Shaping the digital twin for design and production engineering, ”CIRP Annals, vol. 66, no. 1, pp. 141–144, 2017

    work page 2017

  3. [3]

    Fog com- puting: Principles, architectures, and applications,

    A. V. Dastjerdi, H. Gupta, R. N. Calheiros, S. K. Ghosh, and R. Buyya, “Fog com- puting: Principles, architectures, and applications, ” inInternet of things, pp. 61–75, Elsevier, 2016

    work page 2016

  4. [4]

    A survey of fog computing: concepts, applications and issues,

    S. Yi, C. Li, and Q. Li, “A survey of fog computing: concepts, applications and issues, ” inProceedings of the 2015 workshop on mobile big data , pp. 37–42, ACM, 2015

    work page 2015

  5. [5]

    Edge computing: Vision and challenges,

    W. Shi, J. Cao, Q. Zhang, Y. Li, and L. Xu, “Edge computing: Vision and challenges, ” IEEE Internet of Things Journal , vol. 3, no. 5, pp. 637–646, 2016

    work page 2016

  6. [6]

    Fog Computing: Focusing on Mobile Users at the Edge

    T. H. Luan, L. Gao, Z. Li, Y. Xiang, G. Wei, and L. Sun, “Fog computing: Focusing on mobile users at the edge, ”arXiv preprint arXiv:1502.01815, 2015

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  7. [7]

    Fog computing architecture to enable con- sumer centric internet of things services,

    S. K. Datta, C. Bonnet, and J. Haerri, “Fog computing architecture to enable con- sumer centric internet of things services, ” in2015 International Symposium on Con- sumer Electronics (ISCE), pp. 1–2, IEEE, 2015

    work page 2015

  8. [8]

    Mobile edge comput- ingâĂŤa key technology towards 5g,

    Y. C. Hu, M. Patel, D. Sabella, N. Sprecher, and V. Young, “Mobile edge comput- ingâĂŤa key technology towards 5g, ” ETSI white paper , vol. 11, no. 11, pp. 1–16, 2015

    work page 2015

  9. [9]

    Multi-access edge computing: The driver behind the wheel of 5g-connected cars,

    F. Giust, V. Sciancalepore, D. Sabella, M. C. Filippou, S. Mangiante, W. Featherstone, and D. Munaretto, “Multi-access edge computing: The driver behind the wheel of 5g-connected cars, ”IEEE Communications Standards Magazine, vol. 2, no. 3, pp. 66–73, 2018

    work page 2018

  10. [10]

    Digital Twins: The Convergence of Multimedia Technologies,

    A. El Saddik, “Digital Twins: The Convergence of Multimedia Technologies, ”IEEE MultiMedia, vol. 25, pp. 87–92, Apr. 2018

    work page 2018

  11. [11]

    Mind + Machine: What is a Digital Twin?

    GE, “Mind + Machine: What is a Digital Twin?. ” https://www.ge.com/digital/ applications/digital-twin, 2019. [Online] Accessed: 18-May-2019

    work page 2019

  12. [12]

    Representing Industrial Data Streams in Digital Twins Using Semantic Labeling,

    P. Zehnder and D. Riemer, “Representing Industrial Data Streams in Digital Twins Using Semantic Labeling, ” inProc. IEEE International Conference on Big Data (Big Data), (Seattle, WA, USA), pp. 4223–4226, Dec. 2018

    work page 2018

  13. [13]

    Digital Twin: Mitigating Unpredictable, Undesirable Emergent Behavior in Complex Systems,

    M. Grieves and J. Vickers, “Digital Twin: Mitigating Unpredictable, Undesirable Emergent Behavior in Complex Systems, ” inTransdisciplinary perspectives on complex systems, pp. 85–113, Springer, 2017

    work page 2017

  14. [14]

    Digital twin-driven product design, manufacturing and service with big data,

    F. Tao, J. Cheng, Q. Qi, M. Zhang, H. Zhang, and F. Sui, “Digital twin-driven product design, manufacturing and service with big data, ”The International Journal of Advanced Manufacturing Technology, vol. 94, no. 9-12, pp. 3563–3576, 2018

    work page 2018

  15. [15]

    Digital twin: Mitigating unpredictable, undesirable emergent behavior in complex systems,

    M. Grieves and J. Vickers, “Digital twin: Mitigating unpredictable, undesirable emergent behavior in complex systems, ” inTransdisciplinary perspectives on complex systems, pp. 85–113, Springer, 2017

    work page 2017

  16. [16]

    The digital twin paradigm for future nasa and us air force vehicles,

    E. Glaessgen and D. Stargel, “The digital twin paradigm for future nasa and us air force vehicles, ” in53rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference 20th AIAA/ASME/AHS Adaptive Structures Conference 14th AIAA, p. 1818, 2012

    work page 2012

  17. [17]

    Gelernter, Mirror worlds: Or the day software puts the universe in a shoebox

    D. Gelernter, Mirror worlds: Or the day software puts the universe in a shoebox... How it will happen and what it will mean . Oxford University Press, 1993

    work page 1993

  18. [18]

    Multimedia mashups for mirror worlds,

    D. Tilden, A. Singh, N. F. Polys, and P. Sforza, “Multimedia mashups for mirror worlds, ” inProceedings of the 16th International Conference on 3D Web Technology , pp. 155–164, ACM, 2011

    work page 2011

  19. [19]

    Vehicle-to-vehicle wireless communication protocols for enhancing highway traffic safety,

    S. Biswas, R. Tatchikou, and F. Dion, “Vehicle-to-vehicle wireless communication protocols for enhancing highway traffic safety, ” IEEE communications magazine , vol. 44, no. 1, pp. 74–82, 2006

    work page 2006

  20. [20]

    Driving behaviors: Models and challenges for non-lane based mixed traffic,

    G. Asaithambi, V. Kanagaraj, and T. Toledo, “Driving behaviors: Models and challenges for non-lane based mixed traffic, ”Transportation in Developing Economies , vol. 2, no. 2, p. 19, 2016

    work page 2016

  21. [21]

    An approach to design of feedback systems with time delay,

    K. Natori and K. Ohnishi, “An approach to design of feedback systems with time delay, ” in31st Annual Conference of IEEE Industrial Electronics Society, 2005. IECON 2005., p. 6 pp., IEEE

    work page 2005

  22. [22]

    Implementing fault-tolerant services using the state machine approach: A tutorial,

    F. B. Schneider, “Implementing fault-tolerant services using the state machine approach: A tutorial, ”ACM Computing Surveys (CSUR) , vol. 22, no. 4, pp. 299–319, 1990

    work page 1990

  23. [23]

    Convolutional social pooling for vehicle trajectory prediction,

    N. Deo and M. M. Trivedi, “Convolutional social pooling for vehicle trajectory prediction, ” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, pp. 1468–1476, 2018

    work page 2018

  24. [24]

    The highd dataset: A drone dataset of naturalistic vehicle trajectories on german highways for validation of highly automated driving systems,

    R. Krajewski, J. Bock, L. Kloeker, and L. Eckstein, “The highd dataset: A drone dataset of naturalistic vehicle trajectories on german highways for validation of highly automated driving systems, ” in2018 IEEE 21st International Conference on Intelligent Transportation Systems (ITSC) , 2018

    work page 2018

  25. [25]

    Xgboost: A scalable tree boosting system,

    T. Chen and C. Guestrin, “Xgboost: A scalable tree boosting system, ” inProceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, pp. 785–794, ACM, 2016

    work page 2016

  26. [26]

    PI-Edge: A Low-Power Edge Computing System for Real-Time Autonomous Driving Services

    J. Tang, S. Liu, B. Yu, and W. Shi, “Pi-edge: A low-power edge computing system for real-time autonomous driving services, ”arXiv preprint arXiv:1901.04978, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 1901

  27. [27]

    Design of distributed cyber– physical systems for connected and automated vehicles with implementing method- ologies,

    Y. Feng, B. Hu, H. Hao, Y. Gao, Z. Li, and J. Tan, “Design of distributed cyber– physical systems for connected and automated vehicles with implementing method- ologies, ”IEEE Transactions on Industrial Informatics , vol. 14, no. 9, pp. 4200–4211, 2018

    work page 2018

  28. [28]

    Infrastructure enabled autonomy: A distributed intelligence architecture for autonomous vehicles,

    S. Gopalswamy and S. Rathinam, “Infrastructure enabled autonomy: A distributed intelligence architecture for autonomous vehicles, ” in2018 IEEE Intelligent Vehicles Symposium (IV), pp. 986–992, IEEE, 2018

    work page 2018

  29. [29]

    SDN-Based Resource Management for Autonomous Vehicular Networks: A Multi-Access Edge Computing Approach

    H. Peng, Q. Ye, and X. Shen, “Sdn-based resource management for autonomous vehicular networks: A multi-access edge computing approach, ” arXiv preprint arXiv:1809.08966, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  30. [30]

    A novel digital twin-centric approach for driver intention prediction and traffic congestion avoidance,

    S. A. Kumar, R. Madhumathi, P. R. Chelliah, L. Tao, and S. Wang, “A novel digital twin-centric approach for driver intention prediction and traffic congestion avoidance, ”Journal of Reliable Intelligent Environments , vol. 4, no. 4, pp. 199–209, 2018

    work page 2018

  31. [31]

    Digital behavioral twins for safe connected cars,

    X. Chen, E. Kang, S. Shiraishi, V. M. Preciado, and Z. Jiang, “Digital behavioral twins for safe connected cars, ” in Proceedings of the 21th ACM/IEEE International Conference on Model Driven Engineering Languages and Systems , pp. 144–153, ACM, 2018

    work page 2018

  32. [32]

    An Empirical Evaluation of Deep Learning on Highway Driving

    B. Huval, T. Wang, S. Tandon, J. Kiske, W. Song, J. Pazhayampallil, M. Andriluka, P. Rajpurkar, T. Migimatsu, R. Cheng-Yue,et al., “An empirical evaluation of deep learning on highway driving, ”arXiv preprint arXiv:1504.01716, 2015

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  33. [33]

    A survey on motion prediction and risk assessment for intelligent vehicles,

    S. Lefèvre, D. Vasquez, and C. Laugier, “A survey on motion prediction and risk assessment for intelligent vehicles, ”ROBOMECH journal, vol. 1, no. 1, p. 1, 2014

    work page 2014

  34. [34]

    Multi-modal trajectory prediction of surrounding vehicles with maneuver based lstms,

    N. Deo and M. M. Trivedi, “Multi-modal trajectory prediction of surrounding vehicles with maneuver based lstms, ” in2018 IEEE Intelligent Vehicles Symposium (IV) , pp. 1179–1184, IEEE, 2018

    work page 2018

  35. [35]

    How would surround vehicles move? a unified framework for maneuver classification and motion prediction,

    N. Deo, A. Rangesh, and M. M. Trivedi, “How would surround vehicles move? a unified framework for maneuver classification and motion prediction, ”IEEE Transac- tions on Intelligent Vehicles, vol. 3, no. 2, pp. 129–140, 2018

    work page 2018

  36. [36]

    A programming language and system for heterogeneous cloud of things,

    R. Wenger, X. Zhu, J. Krishnamurthy, and M. Maheswaran, “A programming language and system for heterogeneous cloud of things, ” in2016 IEEE 2nd International Conference on Collaboration and Internet Computing (CIC) , pp. 169–177, IEEE, 2016

    work page 2016

  37. [37]

    A Language for Programming Edge Clouds for Next Generation IoT Applications

    M. Maheswaran, R. Wenger, R. Olaniyan, S. Memon, O. Fadahunsi, and R. Echomgbe, “A language for programming edge clouds for next generation iot applications, ”CoRR, vol. abs/1906.09962, 2019

    work page internal anchor Pith review Pith/arXiv arXiv 1906