pith. sign in

arxiv: 2604.13878 · v2 · submitted 2026-04-15 · 💻 cs.LG

Drowsiness-Aware Adaptive Autonomous Braking System based on Deep Reinforcement Learning for Enhanced Road Safety

Hossem Eddine Hafidi , Elisabetta De Giovanni , Teodoro Montanaro , Ilaria Sergi , Massimo De Vittorio , Luigi Patrono This is my paper

Pith reviewed 2026-05-10 13:29 UTC · model grok-4.3

classification 💻 cs.LG
keywords drowsiness detectiondeep reinforcement learningautonomous brakingECG signalsDQN agentroad safetyCARLA simulation
0
0 comments X

The pith

A deep reinforcement learning braking agent that incorporates real-time ECG drowsiness detection avoids collisions 99.99 percent of the time in simulation under both drowsy and alert conditions.

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

The paper develops an autonomous braking system that adapts its behavior to the driver's drowsiness level using deep reinforcement learning. Drowsiness is identified from ECG heart signals by a recurrent neural network and then supplied to a Double-Dueling DQN agent by representing impairment as a delay before the agent can act. The controller is trained and tested inside the CARLA driving simulator, where it reaches a 99.99 percent success rate at preventing crashes whether the driver is drowsy or not. Readers would care because drowsiness is linked to a sizable share of road accidents and existing driver-assistance systems do not adjust braking decisions to the driver's changing physiological state.

Core claim

The authors show that a Double-Dueling Deep Q-Network agent for autonomous braking, whose state includes a drowsiness indicator inferred from ECG signals via an RNN and in which impairment is modeled as an action delay, achieves a 99.99 percent collision-avoidance success rate in the CARLA simulator for both drowsy and non-drowsy drivers.

What carries the argument

A Double-Dueling Deep Q-Network agent whose observable state incorporates drowsiness inferred from ECG signals by an RNN, with driver impairment represented as a delay in action execution.

If this is right

  • Physiology-aware reinforcement learning controllers can sustain high safety performance across both alert and drowsy driver states.
  • Real-time ECG monitoring can be fused directly with vehicle control policies without requiring separate modules.
  • The same state-augmentation approach supports adaptive safety systems that respond to changing driver conditions.
  • Simulation results indicate that incorporating driver state can improve collision avoidance in high-fidelity environments.

Where Pith is reading between the lines

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

  • The method could be extended to other measurable driver impairments such as distraction or intoxication by adding corresponding state variables.
  • Hybrid human-AI driving systems might use similar detection to decide when to override braking control.
  • Transfer to real vehicles would require validation against actual vehicle dynamics and sensor noise not present in simulation.

Load-bearing premise

Representing drowsiness only as a delay added to the agent's actions inside the simulation is sufficient to capture how tiredness actually changes braking decisions and vehicle behavior.

What would settle it

A physical-vehicle test using drivers whose drowsiness is confirmed by simultaneous ECG recording, in which the observed collision-avoidance rate falls substantially below 99.99 percent, would disprove the reported performance.

Figures

Figures reproduced from arXiv: 2604.13878 by Elisabetta De Giovanni, Hossem Eddine Hafidi, Ilaria Sergi, Luigi Patrono, Massimo De Vittorio, Teodoro Montanaro.

Figure 1
Figure 1. Figure 1: System architecture design for real-world drowsiness-aware brake [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: RR Interval trend over the last capsule of a random DEW and NSRW [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 2
Figure 2. Figure 2: Raw and filtered ECG signals with R-peak detection for both normal [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Feature Importance Comparison Across Methods: RFE, MI, PI, and [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Overview of the proposed DRL framework, showing the interaction between the driver drowsiness detection system (left), the DQN agent (center), and [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Reward structure visualization: green labels indicate positive rewards; red labels indicate penalties based on distance and braking behavior. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Finite state automaton representing the reward structure across the [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Front-mounted radar sensor showing forward field of view and [PITH_FULL_IMAGE:figures/full_fig_p010_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Comparison of minimum and average rewards across training episodes. [PITH_FULL_IMAGE:figures/full_fig_p012_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Analysis of the DDDQN agent’s behavior during 999 episodes [PITH_FULL_IMAGE:figures/full_fig_p013_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Impact of drowsiness on the control behavior of the ego vehicle driven by the agent. (a) illustrates the agent’s control strategy under normal conditions, [PITH_FULL_IMAGE:figures/full_fig_p014_11.png] view at source ↗
read the original abstract

Driver drowsiness significantly impairs the ability to accurately judge safe braking distances and is estimated to contribute to 10%-20% of road accidents in Europe. Traditional driver-assistance systems lack adaptability to real-time physiological states such as drowsiness. This paper proposes a deep reinforcement learning-based autonomous braking system that integrates vehicle dynamics with driver physiological data. Drowsiness is detected from ECG signals using a Recurrent Neural Network (RNN), selected through an extensive benchmark analysis of 2-minute windows with varying segmentation and overlap configurations. The inferred drowsiness state is incorporated into the observable state space of a Double-Dueling Deep Q-Network (DQN) agent, where driver impairment is modeled as an action delay. The system is implemented and evaluated in a high-fidelity CARLA simulation environment. Experimental results show that the proposed agent achieves a 99.99% success rate in avoiding collisions under both drowsy and non-drowsy conditions. These findings demonstrate the effectiveness of physiology-aware control strategies for enhancing adaptive and intelligent driving safety systems.

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 paper proposes a drowsiness-aware autonomous braking system that detects driver drowsiness from ECG signals via an RNN (benchmarked on 2-minute windows), incorporates the state into a Double-Dueling DQN agent by modeling impairment as action delay, and evaluates the agent in the CARLA simulator, claiming a 99.99% collision-avoidance success rate under both drowsy and non-drowsy conditions.

Significance. If the empirical result holds under rigorous validation, the integration of real-time physiological sensing with RL-based control could advance adaptive ADAS design. The benchmark analysis of RNN configurations and the choice of high-fidelity CARLA simulation are positive elements that support reproducibility in simulation-based studies.

major comments (3)
  1. [Abstract] Abstract: the central claim of a 99.99% success rate supplies no baseline comparisons (e.g., standard DQN without drowsiness state), no training or test episode counts, no error bars, and no definition of the success metric or scenario distribution, rendering the contribution of the drowsiness-aware component impossible to assess.
  2. [Methodology] Methodology (DQN state-space construction): modeling drowsiness solely as a (presumably fixed) action delay added to the observable state, without introducing stochastic perception noise, altered vehicle dynamics, or variable delay drawn from physiological distributions, is load-bearing for the claim that the agent reliably avoids collisions under real impairment; the RNN output only augments the state and does not affect the underlying simulator physics.
  3. [Experimental results] Experimental results: no description is given of how ground-truth drowsiness labels were obtained or validated for the ECG data used to train the RNN, which directly affects the reliability of the state input and the reported performance under the drowsy condition.
minor comments (2)
  1. [Abstract] The abstract would be clearer if it explicitly defined the success metric (e.g., fraction of episodes with no collision within a fixed time horizon) and the exact CARLA scenario parameters.
  2. [Methodology] Notation for the DQN components (Double-Dueling) and the precise form of the action-delay augmentation should be formalized with equations for reproducibility.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed comments. We address each major comment point by point below, indicating where we agree and what revisions will be made.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim of a 99.99% success rate supplies no baseline comparisons (e.g., standard DQN without drowsiness state), no training or test episode counts, no error bars, and no definition of the success metric or scenario distribution, rendering the contribution of the drowsiness-aware component impossible to assess.

    Authors: We agree that the abstract lacks these details, which hinders assessment of the drowsiness-aware contribution. In the revised manuscript we will expand the abstract to report baseline results from a standard Double-Dueling DQN without the drowsiness state, specify the number of training and test episodes, include error bars or standard deviations on the success rates, and explicitly define the success metric (collision-free episodes) together with the scenario distribution used in CARLA. revision: yes

  2. Referee: [Methodology] Methodology (DQN state-space construction): modeling drowsiness solely as a (presumably fixed) action delay added to the observable state, without introducing stochastic perception noise, altered vehicle dynamics, or variable delay drawn from physiological distributions, is load-bearing for the claim that the agent reliably avoids collisions under real impairment; the RNN output only augments the state and does not affect the underlying simulator physics.

    Authors: We acknowledge that modeling impairment as a fixed action delay is a controlled simplification that does not alter CARLA physics. The state augmentation nevertheless allows the agent to learn compensatory policies. We will revise the methodology section to justify this choice, discuss its limitations relative to real physiological variability, and add new experiments that sample variable delays from physiological distributions to test robustness. revision: partial

  3. Referee: [Experimental results] Experimental results: no description is given of how ground-truth drowsiness labels were obtained or validated for the ECG data used to train the RNN, which directly affects the reliability of the state input and the reported performance under the drowsy condition.

    Authors: We agree this information is missing and will add it. The revised experimental results section will describe the ECG dataset, the process used to obtain ground-truth labels (expert annotation using standard drowsiness scales), and the validation steps (e.g., cross-validation and benchmark accuracy) performed on the RNN. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation or claims

full rationale

The paper presents an empirical RL implementation (Double-Dueling DQN with RNN-derived drowsiness state and action-delay modeling) evaluated via CARLA simulation runs that produce the reported 99.99% success rate. No equations, parameter fits, or self-citations are shown that reduce this outcome to a tautological input, self-definition, or renamed known result. The success metric is an observed simulation statistic rather than a derived quantity forced by construction, leaving the chain self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Based on the abstract alone, the system relies on standard DRL components (Double-Dueling DQN, RNN) and the modeling choice that drowsiness equals action delay; no explicit free parameters, axioms, or invented entities are stated.

pith-pipeline@v0.9.0 · 5500 in / 1123 out tokens · 36045 ms · 2026-05-10T13:29:14.011769+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

46 extracted references · 46 canonical work pages

  1. [1]

    Drowsy driving–19 states and the district of columbia, 2009–2010,

    Centers for Disease Control and Prevention (CDC), “Drowsy driving–19 states and the district of columbia, 2009–2010,” MMWR Morbidity and Mortality Weekly Report, vol. 61, no. 51-52, pp. 1033–1037, 2013

    work page 2009

  2. [2]

    Physiological measurements for driving drowsiness: A comparative study of multi- modality feature fusion and selection,

    Y . Wu, X. Jiang, Y . Guo, H. Zhu, C. Dai, and W. Chen, “Physiological measurements for driving drowsiness: A comparative study of multi- modality feature fusion and selection,” Computers in Biology and Medicine, vol. 167, p. 107590, 2023

    work page 2023

  3. [3]

    European ncap program developments to address driver distraction, drowsiness and sudden sickness,

    R. Fredriksson, M. G. Lenn ´e, S. van Montfort, and C. Grover, “European ncap program developments to address driver distraction, drowsiness and sudden sickness,”Frontiers in Neuroergonomics, vol. 2, p. 786674, 2021

    work page 2021

  4. [4]

    Human-vehicle interaction to support driver’s situation awareness in automated vehicles: A systematic review,

    M. Capallera, L. Angelini, Q. Meteier, O. A. Khaled, and E. Mugellini, “Human-vehicle interaction to support driver’s situation awareness in automated vehicles: A systematic review,” IEEE Transactions on Intel- ligent Vehicles, vol. 8, no. 3, pp. 2551–2567, 2023. 15

    work page 2023

  5. [5]

    Sensor applications and physiological features in drivers’ drowsiness detection: A review,

    A. Chowdhury, R. Shankaran, M. Kavakli, and M. Haque, “Sensor applications and physiological features in drivers’ drowsiness detection: A review,” IEEE Sensors Journal , vol. 18, pp. 3055–3067, 2018

    work page 2018

  6. [6]

    Rethinking sleep analysis,

    H. Schulz, “Rethinking sleep analysis,” Journal of Clinical Sleep Medicine, vol. 4, no. 2, pp. 99–103, 2008

    work page 2008

  7. [7]

    PERCLOS-based technologies for detecting drowsiness: current evidence and future directions,

    T. Abe, “PERCLOS-based technologies for detecting drowsiness: current evidence and future directions,” SLEEP Advances , vol. 4, no. 1, p. zpad006, 01 2023

    work page 2023

  8. [8]

    A survey on state-of-the-art drowsiness detection techniques,

    M. Ramzan, H. Khan, S. M. Awan, A. Ismail, M. Ilyas, and A. Mah- mood, “A survey on state-of-the-art drowsiness detection techniques,” IEEE Access, vol. 7, pp. 61 904–61 919, 2019

    work page 2019

  9. [9]

    Design and implemen- tation of a drowsiness-fatigue-detection system based on wearable smart glasses to increase road safety,

    W.-J. Chang, L.-B. Chen, and Y .-Z. Chiou, “Design and implemen- tation of a drowsiness-fatigue-detection system based on wearable smart glasses to increase road safety,” IEEE Transactions on Consumer Electronics, vol. 64, no. 4, pp. 461–469, 2018

    work page 2018

  10. [10]

    A real-time and lightweight driver fatigue detection model using anchor-free and visual-attention mechanisms,

    J. Wang, B. Li, Z. Li, P. Xu, and L. Li, “A real-time and lightweight driver fatigue detection model using anchor-free and visual-attention mechanisms,” Applied Intelligence , vol. 54, no. 20, pp. 9811–9829, 2024

    work page 2024

  11. [11]

    A. M. Strijkstra, D. G. Beersma, B. Drayer, N. Halbesma, and S. Daan, “Subjective sleepiness correlates negatively with global alpha (8–12 hz) and positively with central frontal theta (4–8 hz) frequencies in the human resting awake electroencephalogram,” Neuroscience Letters, vol. 340, no. 1, pp. 17—-20, Apr 2003

    work page 2003

  12. [12]

    Wearable electroencephalography,

    A. Casson, D. Yates, S. J. M. Smith, J. S. Duncan, and E. Rodr ´ıguez- Villegas, “Wearable electroencephalography,” IEEE Engineering in Medicine and Biology Magazine , vol. 29, pp. 44–56, 2010

    work page 2010

  13. [13]

    Driver drowsiness detection using r-r interval of electrocardiogram and self-attention au- toencoder,

    K. Fujiwara, H. Iwamoto, K. Hori, and M. Kano, “Driver drowsiness detection using r-r interval of electrocardiogram and self-attention au- toencoder,” IEEE Transactions on Intelligent Vehicles, vol. 9, no. 1, pp. 2956—-2965, 2024

    work page 2024

  14. [14]

    A hybrid approach to detect driver drowsiness utilizing physiological signals to improve system performance and wearability,

    M. Awais, N. Badruddin, and M. Drieberg, “A hybrid approach to detect driver drowsiness utilizing physiological signals to improve system performance and wearability,” Sensors (Switzerland) , vol. 17, no. 9, 2017

    work page 2017

  15. [15]

    Heart rate variability: standards of measurement, physiological interpretation and clinical use,

    T. F. of the European Society of Cardiology, the North American Soci- ety of Pacing, and Electrophysiology, “Heart rate variability: standards of measurement, physiological interpretation and clinical use,” Circulation, vol. 93, no. 5, pp. 1043–1065, 1996

    work page 1996

  16. [16]

    Drowsiness detection using radial basis function network with electrocardiographic rr interval statistical feature,

    O. Maftukhaturrizqoh, N. Nuryani, and D. Darmanto, “Drowsiness detection using radial basis function network with electrocardiographic rr interval statistical feature,”Journal of Physics: Conference Series, vol. 1153, no. 1, p. 012049, feb 2019

    work page 2019

  17. [17]

    Ecg based driver drowsiness detection using scalograms and convolutional neural networks,

    A. R. Rachamalla and C. S. Kumar, “Ecg based driver drowsiness detection using scalograms and convolutional neural networks,” AIP Conference Proceedings, vol. 2725, no. 1, p. 020014, 04 2023

    work page 2023

  18. [18]

    Driver drowsiness classification using data fusion of vehicle- based measures and ecg signals,

    S. Arefnezhad, A. Eichberger, M. Fr ¨uhwirth, C. Kaufmann, and M. Moser, “Driver drowsiness classification using data fusion of vehicle- based measures and ecg signals,” in2020 IEEE International Conference on Systems, Man, and Cybernetics (SMC) . IEEE, 2020, pp. 451–456

    work page 2020

  19. [19]

    Detection of driver drowsiness level using a hybrid learning model based on ecg signals,

    H. Xiong, Y . Yan, L. Sun, J. Liu, Y . Han, and Y . Xu, “Detection of driver drowsiness level using a hybrid learning model based on ecg signals,” Biomedizinische Technik. Biomedical Engineering , vol. 69, no. 2, pp. 151–165, 2024

    work page 2024

  20. [20]

    Drowsy driving detection based on fused data and information granulation,

    Y . Wang, L. Jin, K. Li, B. Guo, Z. Yi, and S. Jian, “Drowsy driving detection based on fused data and information granulation,” IEEE Access, vol. 7, pp. 183 739–183 750, 2019

    work page 2019

  21. [21]

    Real-time personalized atrial fibrillation prediction on multi-core wearable sensors,

    E. De Giovanni, A. A. Vald ´es, M. Pe ´on-Quir´os, A. Aminifar, and D. Atienza, “Real-time personalized atrial fibrillation prediction on multi-core wearable sensors,” IEEE Transactions on Emerging Topics in Computing, vol. 9, no. 4, pp. 1654–1666, 2021

    work page 2021

  22. [22]

    Human-level control through deep reinforcement learning,

    V . Mnih, K. Kavukcuoglu, D. Silver, A. A. Rusu, J. Veness, M. G. Bellemare, A. Graves, M. A. Riedmiller, A. K. Fidjeland, G. Ostrovski, S. Petersen, C. Beattie, A. Sadik, I. Antonoglou, H. King, D. Kumaran, D. Wierstra, S. Legg, and D. Hassabis, “Human-level control through deep reinforcement learning,” Nature, vol. 518, no. 7540, pp. 529–533, 2015

    work page 2015

  23. [23]

    Prioritized experience replay-based deep q learning: Multiple-reward architecture for highway driving decision making,

    W. Yuan, Y . Li, H. Zhuang, C. Wang, and M. Yang, “Prioritized experience replay-based deep q learning: Multiple-reward architecture for highway driving decision making,” IEEE Robotics & Automation Magazine, vol. 28, no. 4, pp. 21–31, 2021

    work page 2021

  24. [24]

    On the classification of grbs and their occurrence rates,

    R. Ruffini, J. A. Rueda, M. Muccino, Y . Aimuratov, L. M. Becerra, C. L. Bianco, M. Kovacevic, R. Moradi, F. G. Oliveira, G. B. Pisani, and Y . Wang, “On the classification of grbs and their occurrence rates,” The Astrophysical Journal , vol. 832, no. 2, p. 136, Nov. 2016

    work page 2016

  25. [25]

    Deep reinforcement learning with double q-learning,

    H. van Hasselt, A. Guez, and D. Silver, “Deep reinforcement learning with double q-learning,” in Proceedings of the Thirtieth AAAI Confer- ence on Artificial Intelligence (AAAI’16). AAAI Press, 2016, pp. 2094– 2100

    work page 2016

  26. [26]

    Dueling network architectures for deep reinforcement learning,

    Z. Wang, T. Schaul, M. Hessel, H. Van Hasselt, M. Lanctot, and N. De Freitas, “Dueling network architectures for deep reinforcement learning,” in Proceedings of the 33rd International Conference on Inter- national Conference on Machine Learning - Volume 48 , ser. ICML’16. JMLR.org, 2016, pp. 1995—-2003

    work page 2016

  27. [27]

    Autonomous braking system via deep reinforcement learning,

    H. Chae, C. M. Kang, B. Kim, J. Kim, C. C. Chung, and J. W. Choi, “Autonomous braking system via deep reinforcement learning,” in 2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC), 2017, pp. 1–6

    work page 2017

  28. [28]

    Enhancing the fuel-economy of v2i-assisted autonomous driving: A reinforcement learning approach,

    X. Liu, Y . Liu, Y . Chen, and L. Hanzo, “Enhancing the fuel-economy of v2i-assisted autonomous driving: A reinforcement learning approach,” IEEE Transactions on Vehicular Technology , vol. 69, no. 8, pp. 8329– 8342, 2020

    work page 2020

  29. [29]

    Decision- making strategy on highway for autonomous vehicles using deep rein- forcement learning,

    J. Liao, T. Liu, X. Tang, X. Mu, B. Huang, and D. Cao, “Decision- making strategy on highway for autonomous vehicles using deep rein- forcement learning,” IEEE Access, vol. 8, pp. 177 804–177 814, 2020

    work page 2020

  30. [30]

    Combining planning and deep reinforcement learning in tactical decision making for autonomous driving,

    C.-J. Hoel, K. Driggs-Campbell, K. Wolff, L. Laine, and M. J. Kochen- derfer, “Combining planning and deep reinforcement learning in tactical decision making for autonomous driving,” IEEE Transactions on Intel- ligent Vehicles, vol. 5, no. 2, pp. 294–305, 2020

    work page 2020

  31. [31]

    Navigating occluded intersections with autonomous vehicles using deep reinforcement learning,

    D. Isele, R. Rahimi, A. Cosgun, K. Subramanian, and K. Fujimura, “Navigating occluded intersections with autonomous vehicles using deep reinforcement learning,” in Proceedings of the 2018 IEEE International Conference on Robotics and Automation (ICRA 2018) . Brisbane, Australia: IEEE, 2018, pp. 2034–2039

    work page 2018

  32. [32]

    Deep reinforcement- learning-based driving policy for autonomous road vehicles,

    K. Makantasis, M. Kontorinaki, and I. Nikolos, “Deep reinforcement- learning-based driving policy for autonomous road vehicles,” IET Intel- ligent Transport Systems, vol. 14, no. 1, pp. 13–24, 2020

    work page 2020

  33. [33]

    Enhancing driver attention and road safety through eeg-informed deep reinforcement learning and soft computing,

    M. Yousaf, M. Farhan, Y . Saeed, M. J. Iqbal, F. Ullah, and G. Srivastava, “Enhancing driver attention and road safety through eeg-informed deep reinforcement learning and soft computing,” Applied Soft Computing , vol. 167, p. 112320, 2024

    work page 2024

  34. [34]

    An improved adaptive radar signal sorting algorithm based on dbscan by a novel cvi,

    Y . Su, Z. Chen, L. Gong, X. Xu, and Y . Yao, “An improved adaptive radar signal sorting algorithm based on dbscan by a novel cvi,” IEEE Access, vol. 12, pp. 43 139—-43 154, 2024

    work page 2024

  35. [35]

    Neurocognitive consequences of sleep deprivation,

    N. Goel, H. Rao, J. S. Durmer, and D. F. Dinges, “Neurocognitive consequences of sleep deprivation,” Seminars in Neurology , vol. 29, no. 4, pp. 320–339, Sep 2009

    work page 2009

  36. [36]

    Headway on urban streets: observational data and an intervention to decrease tailgating,

    P. G. Michael, F. C. Leeming, and W. O. Dwyer, “Headway on urban streets: observational data and an intervention to decrease tailgating,” Transportation Research Part F: Traffic Psychology and Behaviour , vol. 3, no. 2, pp. 55–64, 2000

    work page 2000

  37. [37]

    Driver fatigue detection using heart rate variability features from 2-minute electrocardiogram signals while accounting for sex differences,

    C. Zeng, J. Zhang, Y . Su, S. Li, Z. Wang, Q. Li, and W. Wang, “Driver fatigue detection using heart rate variability features from 2-minute electrocardiogram signals while accounting for sex differences,”Sensors, vol. 24, no. 13, p. 4316, 2024

    work page 2024

  38. [38]

    A systematic review of physiological signals based driver drowsiness detection systems,

    U. Saleem, M. I. Butt, A. Ali, and A. Rehman, “A systematic review of physiological signals based driver drowsiness detection systems,” Cognitive Computation and Systems , vol. 15, pp. 101—-118, 2023

    work page 2023

  39. [39]

    NeuroKit2: A python toolbox for neurophysiological signal processing,

    D. Makowski, T. Pham, Z. J. Lau, J. C. Brammer, F. Lespinasse, H. Pham, C. Sch ¨olzel, and S. H. A. Chen, “NeuroKit2: A python toolbox for neurophysiological signal processing,” Behavior Research Methods, vol. 53, no. 4, pp. 1689–1696, feb 2021. [Online]. Available: https://doi.org/10.3758%2Fs13428-020-01516-y

    work page 2021

  40. [40]

    CARLA: An open urban driving simulator,

    A. Dosovitskiy, G. Ros, F. Codevilla, A. Lopez, and V . Koltun, “CARLA: An open urban driving simulator,” in Proceedings of the 1st Annual Conference on Robot Learning , 2017, pp. 1–16

    work page 2017

  41. [41]

    Drivers Drowsiness Database: A collection of physiological signals during the use of a driving simulator (DD- Database),

    L. Orosco, M. A. Garc ´es, G. E. C. Fragapane, C. Dell’Aquila, J. C. I. Gimeno, and E. L. Leber, “Drivers Drowsiness Database: A collection of physiological signals during the use of a driving simulator (DD- Database),” https://doi.org/10.5061/dryad.5tb2rbp9c, 2023, dataset pub- lished on Dryad

    work page doi:10.5061/dryad.5tb2rbp9c 2023

  42. [42]

    Driver behavior and situation aware brake assistance for intelligent vehicles,

    J. McCall and M. Trivedi, “Driver behavior and situation aware brake assistance for intelligent vehicles,” Proceedings of the IEEE , vol. 95, pp. 374–387, 2007

    work page 2007

  43. [43]

    Adaptive brake by wire: From human factors to adaptive implementation,

    A. Spadoni, “Adaptive brake by wire: From human factors to adaptive implementation,” PhD dissertation, University of Trento, 2013

    work page 2013

  44. [44]

    Modular design and optimization of biomedical applications for ultralow power hetero- geneous platforms,

    E. De Giovanni, F. Montagna, B. W. Denkinger, S. Machetti, M. Pe ´on- Quir´os, S. Benatti, D. Rossi, L. Benini, and D. Atienza, “Modular design and optimization of biomedical applications for ultralow power hetero- geneous platforms,” IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems , vol. 39, no. 11, pp. 3821–3832, 2020

    work page 2020

  45. [45]

    A decision- making strategy for vehicle autonomous braking in emergency via deep 16 reinforcement learning,

    Y . Fu, C. Li, F. R. Yu, T. H. Luan, and Y . Zhang, “A decision- making strategy for vehicle autonomous braking in emergency via deep 16 reinforcement learning,” IEEE Transactions on Vehicular Technology , vol. 69, no. 6, pp. 5876–5888, 2020

    work page 2020

  46. [46]

    Towards robust decision-making for autonomous driving on highway,

    K. Yang, X. Tang, S. Qiu, S. Jin, Z. Wei, and H. Wang, “Towards robust decision-making for autonomous driving on highway,” IEEE Transactions on Vehicular Technology, vol. 72, no. 9, pp. 11 251–11 263, 2023. Hossem Eddine Hafidi is a PhD student jointly affiliated with the Istituto Italiano di Tecnologia (IIT) and the IDentification Automation Laboratory (...

    work page 2023