pith. machine review for the scientific record. sign in

arxiv: 2605.07370 · v1 · submitted 2026-05-08 · 💻 cs.RO · cs.AI· cs.MA· cs.SY· eess.SY

Recognition: 2 theorem links

· Lean Theorem

MORPH-U: Multi-Objective Resilient Motion Planning for V2X-Enabled Autonomous Driving in High-Uncertainty Environments via Simulation

Shih-Yu Lai
Authors on Pith no claims yet

Pith reviewed 2026-05-11 02:04 UTC · model grok-4.3

classification 💻 cs.RO cs.AIcs.MAcs.SYeess.SY
keywords autonomous drivingV2X communicationmotion planningresilient systemsCARLA simulatormulti-objective optimizationPareto frontiersensor fusion
0
0 comments X

The pith

A V2X-fused motion planner with Pareto tuning and a sensor-veto gate keeps simulated autonomous vehicles safe under message delays, drops, and forgeries.

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

The paper presents MORPH-U, a closed-loop system that fuses LiDAR, radar, and camera data with V2X CAM and DENM messages to maintain an up-to-date Local Dynamic Map for autonomous driving. It triggers Hybrid-A* replanning only when validated hazards or map changes affect the route, while optimizing tracking error, safety margin, responsiveness, and smoothness through multi-objective selection on the Pareto frontier. A lightweight acceptance gate inspired by quorum rules plus onboard sensor checks filters out suspicious V2X triggers before they cause replanning. CARLA experiments in dynamic scenarios demonstrate safety gains from V2X data, clear trade-off control via Pareto points, and full prevention of replanning even under constant false hazard messages. A reader would care because wireless vehicle communications promise earlier hazard detection but introduce risks that could lead to crashes or erratic behavior if not managed carefully.

Core claim

MORPH-U fuses LiDAR/radar/camera with V2X CAM/DENM into a Local Dynamic Map, triggers Hybrid-A* replanning on validated hazards or map changes, exposes planning/control trade-offs via a multi-objective formulation over tracking error, safety margin, responsiveness, and smoothness with Pareto-frontier selection, and adds a Byzantine-inspired gate combining quorum rules with on-board sensor veto to reject faulty V2X triggers; CARLA experiments show V2X-augmented LDM improves safety, Pareto tuning gives controllable accuracy-comfort choices, and the gate blocks replanning under saturated false-DENM injection.

What carries the argument

The lightweight Byzantine-inspired acceptance gate that combines a quorum rule with an on-board sensor veto to validate V2X hazard messages before triggering replanning.

If this is right

  • V2X-augmented Local Dynamic Maps improve downstream safety in dynamic driving scenarios.
  • Pareto-frontier analysis enables controllable trade-offs between tracking accuracy and driving comfort.
  • The acceptance gate prevents replanning even under maximum-rate false hazard messages.
  • Event-driven replanning stays feasible under real-time constraints when protected by the validation gate.

Where Pith is reading between the lines

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

  • The gate mechanism could reduce reliance on overly conservative default behaviors when V2X reliability is low in actual deployments.
  • Adding energy consumption as a fourth objective in the Pareto analysis might help optimize for battery range in electric vehicles.
  • Similar quorum-plus-sensor checks could protect other connected robotic systems against compromised external data feeds.
  • Varying the fraction of V2X-equipped surrounding vehicles in further simulations would identify the minimum density needed for measurable safety benefits.

Load-bearing premise

The CARLA simulator and its V2X message models accurately capture real-world delays, drops, forgery attacks, and sensor fusion performance so that the observed safety gains and gate effectiveness will transfer outside simulation.

What would settle it

Running the MORPH-U stack on physical test vehicles with real V2X hardware under controlled false message injection and checking whether collision rates and replanning behavior match the simulated results.

Figures

Figures reproduced from arXiv: 2605.07370 by Shih-Yu Lai.

Figure 1
Figure 1. Figure 1: CARLA simulator evaluation setups in MORPH-U. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 5
Figure 5. Figure 5: Hybrid-A* runtime distribution per tick. The tail remains within the [PITH_FULL_IMAGE:figures/full_fig_p004_5.png] view at source ↗
Figure 4
Figure 4. Figure 4: Pareto frontier (Sec. V-D): Tracking vs. Smoothness ( [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
read the original abstract

V2X can warn an autonomous vehicle about hazards beyond line-of-sight, but it also brings uncertainty: messages may be delayed, dropped, or even forged. Meanwhile, map knowledge may change during a trip, forcing the vehicle to replan under tight real-time budgets. This paper studies how to make motion planning and low-level control robust to such uncertain, event-driven updates. We present MORPH-U, a CARLA-based closed-loop stack that fuses LiDAR/radar/camera with V2X (CAM/DENM) into a Local Dynamic Map (LDM) and triggers Hybrid-A* replanning when validated hazards or map changes affect the planned route. We expose the planning/control trade-offs via a multi-objective formulation over tracking error, safety margin (minimum TTC), responsiveness, and smoothness, and select operating points using Pareto-frontier analysis. To avoid unsafe replanning from faulty V2X triggers, MORPH-U adds a lightweight Byzantine-inspired acceptance gate that combines a quorum rule with an on-board sensor veto. Experiments in dynamic CARLA scenarios show that V2X-augmented LDM improves downstream safety, Pareto tuning provides controllable accuracy-comfort trade-offs, and the gate prevents replanning under saturated false-DENM injection ($p_{\text{attack}}=1.0$).

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

3 major / 2 minor

Summary. The manuscript presents MORPH-U, a CARLA-based closed-loop simulation framework for multi-objective resilient motion planning in V2X-enabled autonomous vehicles. It integrates LiDAR/radar/camera data with V2X (CAM/DENM) messages into a Local Dynamic Map (LDM), triggers Hybrid-A* replanning on validated hazards or map changes, formulates planning/control trade-offs over tracking error, minimum TTC, responsiveness and smoothness, selects operating points via Pareto-frontier analysis, and adds a quorum-plus-sensor-veto gate to filter faulty V2X triggers. The central empirical claims are that V2X-augmented LDM improves downstream safety, Pareto tuning yields controllable accuracy-comfort trade-offs, and the gate prevents replanning under saturated false-DENM injection at p_attack=1.0 in dynamic scenarios.

Significance. If the simulation results are reproducible and the gate mechanism proves robust, the work offers a concrete example of multi-objective Pareto tuning and Byzantine-inspired filtering for handling V2X uncertainty in motion planning. The closed-loop CARLA evaluation and explicit trade-off analysis are strengths that could inform future resilient AV stacks, though the simulation-only nature limits immediate practical significance without further validation.

major comments (3)
  1. [Abstract] Abstract: the abstract asserts specific experimental outcomes (safety improvement, gate effectiveness at p_attack=1.0) but supplies no quantitative metrics, error bars, scenario counts, or statistical analysis. This absence makes it impossible to evaluate the magnitude or reliability of the claimed benefits and is load-bearing for the central empirical claims.
  2. [Experiments] Experiments section: all quantitative evidence is generated inside CARLA's V2X message and sensor-fusion models with no calibration against real IEEE 802.11p/5G-V2X traces, hardware-in-the-loop delays, or field-collected false-DENM behavior. If the simulator under-models channel congestion, sensor-veto latency, or fusion error under attack, both the reported TTC gains and the gate's claimed robustness at p_attack=1.0 become simulation artifacts; this transferability issue directly affects the soundness of the safety and resilience conclusions.
  3. [System Architecture / Gate Design] Gate description: the quorum-plus-sensor-veto acceptance gate is described as lightweight and Byzantine-inspired, yet the manuscript does not specify the exact quorum threshold, veto latency relative to LDM update rate, or how false-positive/negative rates are measured. Without these parameters, the claim that the gate blocks replanning at p_attack=1.0 cannot be independently assessed.
minor comments (2)
  1. [Notation] Notation for p_attack and other V2X parameters should be defined once in a table or early section and used consistently in text and figures.
  2. [Figures] Pareto-frontier plots and safety-metric figures should include multiple runs or confidence intervals to convey variability rather than single-run curves.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed comments on our manuscript. We address each major point below, indicating revisions where they strengthen the presentation of our simulation-based study.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the abstract asserts specific experimental outcomes (safety improvement, gate effectiveness at p_attack=1.0) but supplies no quantitative metrics, error bars, scenario counts, or statistical analysis. This absence makes it impossible to evaluate the magnitude or reliability of the claimed benefits and is load-bearing for the central empirical claims.

    Authors: We agree that the abstract would benefit from explicit quantitative indicators. In the revised version we will insert concise numerical results drawn from the experiments, such as mean minimum-TTC gains, replanning success rates under p_attack=1.0, the number of CARLA scenarios evaluated, and standard-deviation ranges, while remaining within the word limit. revision: yes

  2. Referee: [Experiments] Experiments section: all quantitative evidence is generated inside CARLA's V2X message and sensor-fusion models with no calibration against real IEEE 802.11p/5G-V2X traces, hardware-in-the-loop delays, or field-collected false-DENM behavior. If the simulator under-models channel congestion, sensor-veto latency, or fusion error under attack, both the reported TTC gains and the gate's claimed robustness at p_attack=1.0 become simulation artifacts; this transferability issue directly affects the soundness of the safety and resilience conclusions.

    Authors: The manuscript is framed as a closed-loop simulation study whose purpose is to isolate the effects of the proposed multi-objective planner and acceptance gate under controlled uncertainty. We will expand the limitations paragraph to discuss known CARLA V2X modeling assumptions (e.g., idealized channel congestion and sensor fusion) and cite relevant literature on sim-to-real gaps. Full hardware-in-the-loop or field calibration lies outside the present scope and would constitute a separate experimental campaign; the current results remain valid as an existence proof within the simulator. revision: partial

  3. Referee: [System Architecture / Gate Design] Gate description: the quorum-plus-sensor-veto acceptance gate is described as lightweight and Byzantine-inspired, yet the manuscript does not specify the exact quorum threshold, veto latency relative to LDM update rate, or how false-positive/negative rates are measured. Without these parameters, the claim that the gate blocks replanning at p_attack=1.0 cannot be independently assessed.

    Authors: We thank the referee for highlighting this omission. The revised manuscript will state the concrete parameters: a 3-out-of-5 quorum, veto latency bounded by the 100 ms LDM update cycle, and the exact procedure used to compute false-positive/negative rates via controlled ablation runs. These details, together with pseudocode, will allow independent verification of the reported behavior at p_attack=1.0. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical simulation results with no derivation chain

full rationale

The paper describes a CARLA-based closed-loop system for V2X-augmented motion planning, with claims resting entirely on experimental runs in simulation (safety metrics, Pareto fronts, gate behavior under attack). No equations, parameter fittings, or derivations are present that could reduce to inputs by construction. No self-citations are invoked as load-bearing uniqueness theorems or ansatzes. The work is self-contained as an empirical system description; simulation results are generated independently of any internal definitions or prior author results.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The abstract does not identify any free parameters, axioms, or invented entities; the described system appears to rely on established components (Hybrid-A*, LDM, Pareto optimization, Byzantine quorum) without new postulates.

pith-pipeline@v0.9.0 · 5552 in / 1159 out tokens · 53122 ms · 2026-05-11T02:04:44.846874+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

32 extracted references · 32 canonical work pages

  1. [1]

    V2x- bgn: Camera-based v2x-collaborative 3d object detection with bev global non-maximum suppression,

    C. Zhang, B. Tian, S. Meng, S. Qi, Y . Sun, Y . Ai, and L. Chen, “V2x- bgn: Camera-based v2x-collaborative 3d object detection with bev global non-maximum suppression,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 602–607

    work page 2024

  2. [2]

    Simulation-based performance optimization of v2x collective perception by adaptive object filtering,

    Q. Delooz, A. Festag, A. Vinel, and S. C. Lobo, “Simulation-based performance optimization of v2x collective perception by adaptive object filtering,” in2023 IEEE Intelligent V ehicles Symposium (IV), 2023, pp. 1–8

    work page 2023

  3. [3]

    Seecad: Semantic end-to-end communication for autonomous driving,

    S. Ribouh and A. Hadid, “Seecad: Semantic end-to-end communication for autonomous driving,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 1808–1813

    work page 2024

  4. [4]

    Carla-autoware-bridge: Facilitating autonomous driving research with a unified framework for simulation and module development,

    G. Kaljavesi, T. Kerbl, T. Betz, K. Mitkovskii, and F. Diermeyer, “Carla-autoware-bridge: Facilitating autonomous driving research with a unified framework for simulation and module development,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 224–229

    work page 2024

  5. [5]

    Combined registration and fusion of evidential occupancy grid maps for live digital twins of traffic,

    R. v. Kempen, L. Adrian Heidrich, B. Lampe, T. Woopen, and L. Eckstein, “Combined registration and fusion of evidential occupancy grid maps for live digital twins of traffic,” in2023 IEEE Intelligent V ehicles Symposium (IV), 2023, pp. 1–6

    work page 2023

  6. [6]

    CARLA: An open urban driving simulator,

    A. Dosovitskiy, G. Ros, F. Codevilla, A. Lopez, and V . Koltun, “CARLA: An open urban driving simulator,” inProceedings of the 1st Annual Conference on Robot Learning, ser. Proceedings of Machine Learning Research, S. Levine, V . Vanhoucke, and K. Goldberg, Eds., vol. 78. PMLR, 13–15 Nov 2017, pp. 1–16. [Online]. Available: https://proceedings.mlr.press/v...

    work page 2017

  7. [7]

    Simbusters: Bridging simulation gaps in intelligent vehicles perception,

    A. Justo, J. Araluce, J. Romera, M. Rodriguez-Arozamena, L. Gonz ´alez, and S. D ´ıaz, “Simbusters: Bridging simulation gaps in intelligent vehicles perception,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 2471–2476

    work page 2024

  8. [8]

    A fast and elitist multiobjective genetic algorithm: Nsga-ii,

    K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitist multiobjective genetic algorithm: Nsga-ii,”IEEE transactions on evolutionary computation, vol. 6, no. 2, pp. 182–197, 2002

    work page 2002

  9. [9]

    Performance assessment of multiobjective optimizers: An analysis and review,

    E. Zitzler, L. Thiele, M. Laumanns, C. M. Fonseca, and V . G. da Fonseca, “Performance assessment of multiobjective optimizers: An analysis and review,”IEEE transactions on Evolutionary Computation, vol. 7, no. 2, pp. 117–132, 2003

    work page 2003

  10. [10]

    The byzantine generals prob- lem,

    L. Lamport, R. Shostak, and M. Pease, “The byzantine generals prob- lem,”ACM Transactions on Programming Languages and Systems, vol. 4, no. 3, pp. 382–401, 1982

    work page 1982

  11. [11]

    Machine learning with adversaries: Byzantine tolerant gradient descent,

    P. Blanchard, E. M. El Mhamdi, R. Guerraoui, and J. Stainer, “Machine learning with adversaries: Byzantine tolerant gradient descent,” in Advances in Neural Information Processing Systems, 2017

    work page 2017

  12. [12]

    Multi-sensor data fusion using bayesian programming: An automotive application,

    C. Coue, T. Fraichard, P. Bessiere, and E. Mazer, “Multi-sensor data fusion using bayesian programming: An automotive application,” in Intelligent V ehicle Symposium, 2002. IEEE, vol. 2, 2002, pp. 442–447 vol.2

    work page 2002

  13. [13]

    Data fusion using improved dempster- shafer evidence theory for vehicle detection,

    W. Zhao, T. Fang, and Y . Jiang, “Data fusion using improved dempster- shafer evidence theory for vehicle detection,” inF ourth International Conference on Fuzzy Systems and Knowledge Discovery (FSKD 2007), vol. 1, 2007, pp. 487–491

    work page 2007

  14. [14]

    Heterogeneous data fusion for accurate road user tracking: A distributed multi-sensor collaborative approach,

    S. Mentasti, A. Barbiero, and M. Matteucci, “Heterogeneous data fusion for accurate road user tracking: A distributed multi-sensor collaborative approach,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 1658–1665

    work page 2024

  15. [15]

    sustainable

    L.-C. Chen, Y . Zhu, G. Papandreou, F. Schroff, and H. Adam, “Encoder-decoder with atrous separable convolution for semantic image segmentation,”arXiv:1802.02611, 2018

    work page arXiv 2018

  16. [16]

    Pix2planning: End- to-end planning by vision-language model for autonomous driving on carla simulator,

    X. Mu, T. Qin, S. Zhang, C. Xu, and M. Yang, “Pix2planning: End- to-end planning by vision-language model for autonomous driving on carla simulator,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 2383–2390

    work page 2024

  17. [17]

    Blockchain-based vehicular communica- tions: Security and privacy,

    L. Chen, R. Lu, and X. Lin, “Blockchain-based vehicular communica- tions: Security and privacy,”IEEE Network, vol. 31, no. 6, pp. 46–52, 2017

    work page 2017

  18. [18]

    Opendrive 2010 and beyond–status and future of the de facto standard for the description of road networks,

    M. Dupuis, M. Strobl, and H. Grezlikowski, “Opendrive 2010 and beyond–status and future of the de facto standard for the description of road networks,” inProc. of the Driving Simulation Conference Europe, 2010, pp. 231–242

    work page 2010

  19. [19]

    Hd maps: Exploiting opendrive potential for path planning and map monitoring,

    A. Diaz-Diaz, M. Oca ˜na, ´A. Llamazares, C. G ´omez-Hu´elamo, P. Re- venga, and L. M. Bergasa, “Hd maps: Exploiting opendrive potential for path planning and map monitoring,” in2022 IEEE Intelligent V ehicles Symposium (IV), 2022, pp. 1211–1217

    work page 2022

  20. [20]

    Lanelet2: A high-definition map framework for the future of automated driving,

    F. Poggenhans, J.-H. Pauls, J. Janosovits, S. Orf, M. Naumann, F. Kuhnt, and M. Mayr, “Lanelet2: A high-definition map framework for the future of automated driving,” inProc. IEEE Intell. Trans. Syst. Conf., Hawaii, USA, November 2018. [Online]. Available: http:// www.mrt.kit.edu/z/publ/download/2018/Poggenhans2018Lanelet2.pdf

    work page 2018

  21. [21]

    Hd map verification without accurate localization prior using spatio-semantic 1d signals,

    J.-H. Pauls, T. Strauss, C. Hasberg, M. Lauer, and C. Stiller, “Hd map verification without accurate localization prior using spatio-semantic 1d signals,” in2020 IEEE Intelligent V ehicles Symposium (IV), 2020, pp. 680–686

    work page 2020

  22. [22]

    Evaluation of high definition map-based self-localization against occlusions in urban area,

    Y . Endo, E. Javanmardi, Y . Gu, and S. Kamijo, “Evaluation of high definition map-based self-localization against occlusions in urban area,” in2021 IEEE Intelligent V ehicles Symposium (IV), 2021, pp. 920–927

    work page 2021

  23. [23]

    Terminology and analysis of map deviations in urban domains: Towards dependability for hd maps in automated vehicles,

    C. Plachetka, N. Maier, J. Fricke, J.-A. Term ¨ohlen, and T. Fingscheidt, “Terminology and analysis of map deviations in urban domains: Towards dependability for hd maps in automated vehicles,” in2020 IEEE Intelligent V ehicles Symposium (IV), 2020, pp. 63–70

    work page 2020

  24. [24]

    Lanemapnet: Lane network recognization and hd map construction using curve region aware temporal bird’s-eye-view perception,

    T. Zhu, J. Leng, J. Zhong, Z. Zhang, and C. Sun, “Lanemapnet: Lane network recognization and hd map construction using curve region aware temporal bird’s-eye-view perception,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 2168–2175

    work page 2024

  25. [25]

    High integrity lane-level occupancy estimation of road obstacles through lidar and hd map data fusion,

    E. Bernardi, S. Masi, P. Xu, and P. Bonnifait, “High integrity lane-level occupancy estimation of road obstacles through lidar and hd map data fusion,” in2020 IEEE Intelligent V ehicles Symposium (IV), 2020, pp. 1873–1878

    work page 2020

  26. [26]

    Hd map generation from noisy multi-route vehicle fleet data on highways with expectation maximization,

    F. Immel, R. Fehler, M. M. Ghanaat, F. Ries, M. Haueis, and C. Stiller, “Hd map generation from noisy multi-route vehicle fleet data on highways with expectation maximization,” in2023 IEEE Intelligent V ehicles Symposium (IV), 2023, pp. 1–7

    work page 2023

  27. [27]

    E-mlp: Effortless online hd map construction with linear priors,

    R. Li, H. Shan, H. Jiang, J. Xiao, Y . Chang, Y . He, H. Yu, and Y . Ren, “E-mlp: Effortless online hd map construction with linear priors,” in 2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 1008–1014

    work page 2024

  28. [28]

    Smartmot: Exploiting the fusion of hdmaps and multi- object tracking for real-time scene understanding in intelligent vehicles applications,

    C. G ´omez-Hu´elamo, L. M. Bergasa, R. Guti ´errez, J. F. Arango, and A. D ´ıaz, “Smartmot: Exploiting the fusion of hdmaps and multi- object tracking for real-time scene understanding in intelligent vehicles applications,” in2021 IEEE Intelligent V ehicles Symposium (IV), 2021, pp. 710–715

    work page 2021

  29. [29]

    Carlos: An open, modular, and scalable simulation framework for the development and testing of software for c-its,

    C. Geller, B. Haas, A. Kloeker, J. Hermens, B. Lampe, T. Beemel- manns, and L. Eckstein, “Carlos: An open, modular, and scalable simulation framework for the development and testing of software for c-its,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 3100–3106

    work page 2024

  30. [30]

    Emerging of v2x paradigm in the development of a ros-based cooperative architecture for transportation system agents,

    C. M. Elias, O. M. Shehata, E. I. Morgan, and C. Stiller, “Emerging of v2x paradigm in the development of a ros-based cooperative architecture for transportation system agents,” in2022 IEEE Intelligent V ehicles Symposium (IV), 2022, pp. 1303–1308

    work page 2022

  31. [31]

    Co-simulate no more: The carla v2x sensor,

    D. Grimm, M. Schindewolf, D. Kraus, and E. Sax, “Co-simulate no more: The carla v2x sensor,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 2429–2436

    work page 2024

  32. [32]

    Contextualfusion: Context- based multi-sensor fusion for 3d object detection in adverse operating conditions,

    S. Sural, N. Sahu, and R. R. Rajkumar, “Contextualfusion: Context- based multi-sensor fusion for 3d object detection in adverse operating conditions,” in2024 IEEE Intelligent V ehicles Symposium (IV), 2024, pp. 1534–1541

    work page 2024