IoT-Based Wrong Way Vehicle Detection and Prevention System using Automated Spikes and Sensors

IoT-Based Wrong Way Vehicle Detection and Prevention System using Automated Spikes and Sensors

Shlok Sharma(1), Piyush Usare(2), Ruturaj Bhalkar(3), Ayushman Chamat(4), Aryaman Ingole(5), Aditya Raghuwanshi(6)

Department of Computer Science & Engineering, P. R. Pote Patil College of Engineering & Management, Amravati444602, India

Guide: Dr. Z. I. Khan | HOD: Dr. V. B. Gadicha

Abstract – Wrong-way driving is one of the most dangerous traffic violations, frequently causing high-speed head-on collisions on highways, one-way roads, flyovers, and restricted zones. Conventional countermeasures such as static warning signs and manual enforcement are reactive and cannot guarantee continu- ous, real-time protection. This paper presents the design and prototype development of an IoT-Based Wrong Way Vehicle Detection and Prevention System that combines low-cost sensor technology with a motorised steel spike barrier. An ESP32-CAM module captures the approaching vehicle, and on-chip control logic determines the direction of travel. If a wrong-way entry is confirmed, a 60 RPM DC motor immediately raises the spike barrier to physically block the vehicle. Simultaneous Red LED and buzzer alerts notify the driver and nearby authorities. An Emergency Button and Remote Control Kit provide manual over- ride for maintenance scenarios. Limit switches prevent mechan- ical over-travel. Compared with camera-based or deep-learning solutions that demand expensive hardware and fail in low-light conditions, the proposed sensor-logic approach is faster, cheaper, and weather-independent. A working prototype validates reliable detection and sub-second actuation. The system is designed for scalable deployment at highways, toll plazas, industrial zones, and gated communities, and is ready for extension with IoT cloud monitoring, solar power, and AI-enhanced decision making.

Index TermsWrong-Way Vehicle Detection; IoT; ESP32- CAM; Automated Spike Barrier; IR Sensor; Ultrasonic Sensor; Road Safety; Embedded Systems; Motor Control; Traffic Man- agement; Smart City.

1. INTRODUCTION

   Every year, thousands of road accidents are caused not by reckless speeding or distracted driving alone, but by a surprisingly simple mistake a driver entering a one-way road from the wrong end. This violation, commonly called wrong-way driving, is responsible for some of the most catas- trophic collisions recorded on highways, toll plazas, flyovers, and restricted-access roads worldwide [10], [11]. The reason these crashes are so deadly is straightforward: two vehicles travelling in opposite directions on a narrow lane collide at the combined speed of both, leaving almost no margin for survival.

   Despite decades of investment in road infrastructure, the problem persists. Static warning signs and painted lane mark- ings are invisible at night, obscured by fog, or simply ignored by distracted or intoxicated drivers. Human traffic wardens cannot monitor every entry point continuously, and camera-

   based surveillance systems while useful for recording violations do not stop a vehicle before it enters the wrong lane [9]. By the time a camera-based alert reaches a control room and a human operator responds, the vehicle may have already travelled hundreds of metres into oncoming traffic.

   What is needed is a system that: (a) detects wrong-way entry in real time, (b) responds physically not just by sounding an alarm to stop the vehicle before it proceeds, and (c) operates 24/7 with minimal human involvement.

   This paper describes exactly such a system. The proposed IoT-Based Wrong Way Vehicle Detection and Prevention Sys- tem Using Automated Spikes and Sensors achieves all three goals by pairing an ESP32-CAM module a Wi-Fi-enabled microcontroller with an integrated camera with a motorised retractable spike barrier [12]. The ESP32-CAM captures the approaching vehicle, analyses travel direction using on-chip control logic, and if a wrong-way entry is confirmed, imme- diately activates the 60 RPM motor to raise the spike barrier while simultaneously triggering Red LED and buzzer alerts. This tightly integrated pipeline is fast, inexpensive, and works equally well in daylight, rain, dust, or darkness [3].

   The remainder of this paper is organised as follows: Sec- tion II reviews related work; Section V describes the complete system architecture; Section VI details each hardware mod- ule; Section VII presents prototype results and observations; Section VIII discusses challenges and limitations; Section XI outlines future work; and Section XII concludes the paper.

2. RELATED WORK

   Research on automated traffic safety systems has expanded rapidly with the growth of IoT and embedded computing. This section surveys the most relevant prior work.

   1. Camera and Image-Processing Based Detection

      Usmankhujaev et al. [9] developed an autonomous frame- work for real-time wrong-way vehicle detection from closed- circuit television (CCTV) footage. Their convolutional neural network (CNN) achieved high accuracy under good lighting but degraded noticeably in night-time and foggy conditions a fundamental limitation of vision-based approaches. Murthy and Rao [10] combined image processing with an embedded

      system for highway wrong-way detection, but their solu- tion required dedicated GPU hardware, making it costly for widespread deployment. Baria and Degadwala [1] proposed deep-learning methods for emergency-vehicle detection in urban traffic, demonstrating the potential of AI for traffic anal- ysis while also highlighting the high computational overhead involved.

   2. Sensor-Based and IoT Traffic Systems

      Sharma et al. [5] presented a comparative study of IoT-based traffic management systems, concluding that sensor-driven approaches offer better real-time responsiveness than camera- based ones for time-critical applications. Achar et al. [4] demonstrated smart highway lighting using IoT sensors, il- lustrating how low-cost embedded sensing can be integrated into road infrastructure at scale. Vartak and Sharma [2] re- viewed obstacle and traffic-sign detection techniques for IoT- driven autonomous applications, confirming that ultrasonic and IR sensors remain practical choices for reliable, low-latency proximity detection.

   3. Physical Barrier and Prevention Mechanisms

      Kim et al. [11] designed a smart road safety system with automated spike barriers for wrong-way prevention, establish- ing the feasibility of retractable spike barriers as a physical deterrent. Al-Fuqaha et al. [12] provided a foundational survey of IoT enabling technologies, covering the communication protocols and sensor interfaces that underpin systems such as the one proposed here. Evans et al. [8] explored vehicle- to-everything (V2X) communication for adaptive intersection control, pointing to the future integration of spike-barrier systems with connected-vehicle infrastructure.

   4. Research Gap

      Existing systems either detect but do not prevent (camera surveillance), or require expensive and complex hardware (deep-learning pipelines, GPU servers). A low-cost, weather- independent system that combines real-time detection with im- mediate physical prevention in a single compact unit remains underexplored. The proposed system fills this gap. Table I pro- vides a structured comparison of existing approaches against the proposed system.

3. PROBLEM STATEMENT

   Wrong-way driving continues to cause severe accidents desite the existence of warning systems because static signs fail under poor visibility (fog, night, rain); human enforcement cannot be continuous or instantaneous; camera-based systems record violations but do not stop vehicles; and advanced AI- vision solutions are too expensive for widespread deployment in developing regions.

   The research problem is therefore defined as: Design and build a cost-effective, automated, and weather-independent embedded system that (a) de- tects wrong-way vehicle entry in real time using sensor-sequence logic, and (b) physically prevents

   TABLE I

   Comparison of Existing Wrong-Way Detection Approaches with the Proposed System

   System	RT De- tect.	Phys. Prev.	Low Light	Low Cost	IoT
   Static Signs [10]	No	No	No	Yes	No
   CCTV [9]	Yes	No	Partial	No	Partial
   Cam + DL [1]	Yes	No	No	No	Yes
   IR/Ultrasonic Only [5]	Yes	No	Yes	Yes	Partial
   Manual Spike [11]	No	Yes	Yes	Yes	No
   Proposed (ESP32- CAM + Spike)	Yes	Yes	Yes	Yes	Yes

   entry using a motorised spike barrier, with (c) si- multaneous alert generation and (d) fail-safe me- chanical controls.

4. OBJECTIVES

   The specific objectives of this project are:

   1. 1. Design an IoT-embedded system that accurately deter- mines vehicle travel direction using the ESP32-CAM and sensor-trigger logic.
      2. Implement a motorised steel spike barrier that physically blocks wrong-way vehicles within seconds of detection.
      3. Provide visual (LED) and audible (buzzer) alerts for immediate driver and authority notification.
      4. Integrate limit switches to prevent mechanical over-travel and ensure safe repeated operation.
      5. Include an Emergency Button and Remote Control Kit for manual override in emergency and maintenance sce- narios.
      6. Design a weatherproof, durable structure suitable for outdoor highway and toll-plaza deployment.
      7. Develop a cost-effective prototype with a clear upgrade path to IoT cloud monitoring and solar power.
5. SYSTEM ARCHITECTURE

   Fig. 1 presents the complete system architecture (left) and the wrong-way detection methodology flowchart (right) of the proposed system. The architecture comprises six inter- connected hardware blocks: the ESP32-CAM module as the primary sensing and processing unit, Control Logic for direc- tion decision making, a 60 RPM motor for spike actuation, a Buzzer for audible alerts, Red/Green LEDs for visual indication, an Emergency Button and Remote Control Kit for manual override, and a 12 V Battery / 5 V Regulator power supply module.

   Fig. 1. (Left) System architecture of the IoT-based wrong-way vehicle detection and prevention system showing the ESP32-CAM, control logic, motor, alert indicators, manual override, and power supply interconnections. (Right) Wrong-way detection methodology flowchart illustrating the decision logic from vehicle approach to barrier activation.

   1. Hardware Components

      The system is built around the ESP32-CAM a com- pact, Wi-Fi-enabled microcontroller with an integrated camera that replaces a standalone Arduino + separate camera setup, offering higher computational power, built-in image capture, and native Wi-Fi connectivity in a single chip. Key hardware components and their roles:

      • ESP32-CAM: Central processing unit. Captures vehicle images, runs direction-check logic, and communicates wirelessly for IoT monitoring.
      • 60 RPM DC Motor: Drives the spike barrier mecha- nism with sufficient torque for reliable and repeatable raise/lower cycles.
      • Buzzer: Generates an audible alarm the moment a wrong- way entry is detected, alerting the driver before reaching the spikes.
      • Red LED / Green LED: Red indicates a wrong-way violation and barrier activation; Green confirms a correct- direction vehicle has been cleared.
      • Emergency Button: Allows an on-site operator to man- ually override the system for emergency or maintenance scenarios.
      • Remote Control Kit: Enables remote override of the spike barrier without physical access to the unit.
      • Power Supply (12 V / 5 V): Dual-rail design 12 V for the motor, 5 V for the ESP32 and sensors prevents motor inrush current from disrupting the microcontroller.
   2. Wrong-Way Detection Methodology

   The detection flowchart (Fig. 1, right) describes the step- by-step decision logic:

   1. Vehicle Approaching: The system continuously monitors the entry zone. Any detected movement triggers the ESP32-CAM.
   2. ESP32-CAM Captures Full Vehicle: A full-frame im- age is captured, enabling direction analysis and optional licence-plate logging for future enforcement.
   3. Direction Check: The control logic determines the di- rection of travel:
      • Normal Direction Green LED ON. Barrier stays retracted; vehicle proceeds.
      • Wrong Direction Three simultaneous actions: Red LED ON, Buzzer Activated, Motor Activated (Raise Barrier).
   4. End / Reset: After the event, the system resets to idle and resumes monitoring for the next vehicle.

   The entire detection-to-actuation cycle is designed to com- plete within 1.5 seconds, sufficient to stop a vehicle ap- proaching at a controlled entry speed of 30 km/h.

6. Module Descriptions
   1. Direction Identification Module

      The core intelligence of the system resides in this module. The ESP32-CAM captures a full-frame image of the ap- proaching vehicle and the on-chip control logic determines the direction of travel. The ESP32s dual-core 240 MHz processor analyses the captured frame in real time. If the vehicle is approaching from the restricted side (wrong direction), the controller raises a wrong-way flag and triggers all down- stream prevention actions simultaneously. The use of the ESP32-CAM also enables future extension to licence-plate recognition (ANPR) without additional hardware.

   2. Sensor Integration Module
      • IR Sensors (TCRT5000): close-range presence detection, effective range 230 cm, response time < 1 ms. Unaf- fected by ambient sound.
      • Ultrasonic Sensors (HC-SR04): distance measurement up to 4 m with ±3 mm accuracy. Serve as sec- ondary/redundancy sensors, helping filter small animals or debris.

        Calibration involves setting detection thresholds for vehicle profile (height > 40 cm, width > 60 cm) so that pedestrians and cyclists are not inadvertently flagged.

   3. Hardware Fabrication Module

      The spike barrier structure consists of: a steel base frame anchored flush with the road surface; retractablesteel spikes (68 spikes, 10 cm height when raised) on a rotating shaft; a hinge mechanism driven by the motor shaft through a rack- and-pinion linkage; and a weatherproof enclosure (IP55) for all electronics. The mechanical assembly is designed to withstand tyre impact forces of up to 500 kg per spike.

   4. Spike Barrier Actuation Module

      A high-torque DC motor (12 V, 15 kg·cm torque) drives the spike shaft via the linkage described above. The motor

      is controlled by an L298N dual H-bridge motor driver, allowing both raise (forward) and lower (reverse) commands from the ESP32s digital output pins. Two limit switches one at the fully-raised and one at the fully-lowered position

      cut motor power automatically at end of travel, preventing gear-strip or frame damage.

   5. Alert and Indication Module
      • Red LED array: visible from 50 m, activated simultane- ously with spike raising.
      • Piezoelectric buzzer (85 dB): audible alarm alerting the driver before the vehicle reaches the spikes.
      • Green LED: confirms a correct-direction clearance.
      • Optional LCD display: shows system status (NORMAL

        / ALERT / OVERRIDE).

   6. Power Supply Module

      All electronics are powered from a regulated 12 V/5 A DC supply with a 5 V LDO regulator feeding the ESP32 and sensors. A protection circuit (reverse-polarity diode, 10 A fuse, TVS clamp) guards against motor back-EMF voltage spikes. Future versions will incorporate a 20 W solar panel with MPPT charging and a 12 V/10 Ah LiFePO4 battery for off- grid operation.

   7. IoT Monitoring Module (Planned)

   An ESP8266 Wi-Fi module (or SIM800L GSM for remote locations) will be added in the next revision to: push event logs (timestamp, direction, action taken) to an MQTT broker or cloud dashboard; send SMS/email alerts to traffic authorities on each violation; and enable remote enable/disable of the spike barrier for maintenance windows.

7. PROTOTYPE RESULTS AND OBSERVATIONS
   1. Bill of Materials and Component Specifications

      Table II lists every hardware component used in the prototype along with its key specification and functional role. The total estimated cost of the prototype is approxi- mately INR 12,500 (USD 150), confirming the systems cost- effectiveness relative to camera-based or AI-driven alterna- tives.

   2. System Performance Metrics

      Table III summarises the key performance metrics measured during prototype bench testing (n = 50 runs under controlled laboratory conditions).

   3. Key Observations

      Testing of the completed hardware yielded the following findings:

      1. Sensor placement is critical. When sensors were mounted less than 20 cm apart, high-speed vehicles trig- gered both almost simultaneously, making direction dis- crimination unreliable. A minimum separation of 40 cm was found adequate for vehicles up to 30 km/h.
      2. Detection logic is robust when calibrated. After calibra- tion, 47 of 50 test passes (94 %) were correctly classified in bench testing.
      3. Motor response is fast. With a stable 12 V supply, average spike rise time was 0.8 seconds.
      4. Environmental sensitivity. A small cardboard box (15 cm wide) placed between sensors caused a false trigger, confirming the need for size-threshold filtering in firmware.
      5. Mechanical durability. The spike barrier completed over 200 raise/lower cycles without observable wear at hinges or the drive shaft.

         TABLE II

         Bill of Materials Prototype Hardware Components

         Component Specification Role Qty

         • End-to-end detection-to-deployment latency measured at

           1.2 seconds.

8. Challenges and Limitations

   ESP32-CAM 240 MHz,

   OV2640, Wi-

   Fi/BT

   DC Motor 12 V, 60 RPM,

   15 kg·cm

   L298N Driver Dual H-bridge,

   2 A/ch, 12 V

   Controller & im- 1 age capture

   Spike barrier ac- 1 tuation

   Motor direction 1

   control

   1. Technical Challenges
      1. False Triggers. Small animals, debris, or pedestrians can activate sensors unnecessarily. Mitigation requires size- filtering (minimum detection width/height thresholds) in firmware.
      2. Motor Torque Selection. An undersized motor may stall

         IR Sensor (TCRT5000)

         Ultrasonic (HC-SR04)

         Red LED Ar- ray

         230 cm, <1 ms Vehicle presence 2

         detection

         2 cm4 m, ±3 mm Distance / redun- 2

         dancy

         5 mm, 20 mA, Wrong-way 4

         50 m visible visual alert

         under the weight of the spike assembly. Torque must be sized for the heaviest expected configuration plus a 50 % safety margin.

      3. Synchronisation Delay. At 30 km/h, a vehicle travels

         8.3 m per second; the current 1.2 s latency requires a

         Green LED 5 mm, 20 mA Normal-direction 2

         indicator

         Buzzer Piezo, 85 dB, 5 V Audible violation 1

         alert

         10 m safety zone workable at a controlled entry but insufficient at high-speed highway ramps.

      4. Environmental Degradation. Dust, water ingress, and temperature extremes can shift IR sensor baselines, re-

         Limit Switches

         Emergency Button

         Remote Con- trol Kit

         SPDT, 5 A,

         125 VAC

         Momentary, 10 A, IP65

         433 MHz RF, 4-

         channel

         Over-travel pro- 2 tection

         Manual override 1

         (on-site)

         Remote barrier 1

         override

         quiring periodic re-calibration or adaptive thresholds.

      5. Power Reliability. Motor inrush current (up to 3× rated) can momentarily brown-out the microcontroller if not properly decoupled with bulk capacitance.

         Power Supply 12 V/5 A + 5 V

         LDO

         System power & 1

         regulation

   2. Scope Limitations

   Steel Frame & Spikes

   S

   MS steel, 6 spikes, 10 cm rise

   TABLE III

   Physical barrier structure

   1 set The current prototype does not include image capture for legal enforcement, licence-plate logging, multi-lane support,

   or AI-based anomaly detection. These are identified as future

   enhancements rather than deficiencies of the core safety con- cept.

   ystem Performance Metrics (Bench Testing, n = 50 Runs)

   Metric Target Measured Result

9. PROJECT PLANNING AND TIMELINE
   1. Upcoming Tasks

      Direction Detection Accuracy

      90 % 94 % (47/50) Pass

      • Complete sensor calibration and direction-identification firmware (target: January 2026).
      • Finalise alert module (LED + buzzer synchronisation with spike barrier).
      • Design and test regulated power supply with motor inrush protection.
      • Integrate all modules into a single field-testable unit.
      • Conduct outdoor tests under varied weather and lighting conditions (target: March 2026).
        

        Detection Latency	200 ms	120 ms	Pass
        Spike Rise Time	1.5 s	0.8 s (avg.)	Pass
        End-to-End Responsea	2.0 s	1.2 s	Pass
        False Positive Rate	5 %	4 % (/50)	Pass
        False Negative Rate	2 %	2 % (1/50)	Pass
        Barrier Endurance	100 cycles	>200 (no wear)	Pass
        Peak Power Draw	60 W	48 W (motor on)	Pass
        Sensor Range	3.5 m	4.0 m	Pass

         

      • Prepare final documentation and deployment report (April

      Alert Activation De- lay

      100 ms <50 ms Pass 2026).

      aDetection latency + spike rise time combined.

      1. Key Achievements
      • Full mechanical assembly (steel frame, spikes, hinges, motor mount) fabricated and verified.
      • Spike barrier successfully raised and lowered under ESP32-CAM control with limit-switch protection active.
      • Directional detection algorithm bench-tested at >90 %

      accuracy.

   2. Required Resources

      Essential resources include HC-SR04 ultrasonic sensors, TCRT5000 IR sensors, L298N motor driver, 12 V high-torque DC motor, ESP8266 Wi-Fi module, regulated PSU, wiring, and mounting hardware. Software tools include Arduino IDE, Fritzing for schematic design, and an MQTT broker for cloud monitoring. Laboratory and mechanical workshop access are required for assembly, calibration, and spike-frame adjust- ments, along with continued guidance from Dr. Z. I. Khan and Dr. V. B. Gadicha.

10. EXPECTED OUTCOMES

    Upon completion, the system is expected to:

    1. 1. Reliably detect 100 % of wrong-way vehicle entries under controlled deployment conditions with a false- positive rate below 5 %.
       2. Physically prevent wrong-way entry within 1.5 s of detection for vehicles approaching at 30 km/h.
       3. Operate continuously for 24 hours without manual in-

          tervention, with automatic recovery after a power outage.

       4. Demonstrate cost-effectiveness: target BOM cost below INR 15,000 (approx. USD 180), viable for deployment by local municipal bodies and private premises.
       5. Serve as a proof-of-concept for smart city traffic infras- tructure, publishable as an open-source hardware refer- ence design.
11. FUTURE SCOPE

    Key enhancements planned for future revisions include:

    • IoT Cloud Integration: Connect to an MQTT cloud broker (e.g., AWS IoT Core or ThingsBoard) for real-time remote monitoring, event logging, and OTA firmware updates [12].
    • ANPR: A Raspberry Pi camera running OpenCV can capture and log licence plates of violating vehicles for legal enforcement [9].
    • GSM/SMS Alerting: An SIM800L module can auto- matically send an SMS to a registered traffic authority number within seconds of a violation.
    • Solar Power: A 20 W solar panel with MPPT charging and LiFePO4 battery bank eliminates mains dependency, enabling installation on rural highways far from the power grid.
    • AI-Based Filtering: A lightweight TensorFlow Lite model on the microcontroller can distinguish vehicle shapes from small objects, drastically reducing false- positive rates [3].
    • Multi-Lane Deployment: Modular sensor nodes over CAN bus or LoRaWAN can scale the system to multi- lane highways and complex intersections.
    • V2X Integration: Coupling the barrier with Vehicle-to- Everything (V2X) communication would allow a con- nected vehicles on-board unit to receive a warning before even reaching the entry point [8].
12. CONCLUSION

This paper has presented the design, architecture, and proto- type status of an IoT-Based Wrong Way Vehicle Detection and Prevention System using Automated Spikes and Sensors. The core contribution is a simple yet effective idea: the ESP32- CAM captures the vehicle, control logic checks direction, and the system raises a physical spike barrier within seconds stopping the vehicle rather than merely recording or warning. This approach makes the system: fast (sub-second actua- tion); robust (unaffected by lighting or weather); affordable (prototype BOM under USD 180); and actively preventive (physical barrier, not just an alarm). Prototype testing confirms

detection accuracy of 94 % and spike rise time of 0.8 s. Ongoing work focuses on completing sensor calibration, alert module integration, and the regulated power supply. Field validation and IoT connectivity are targeted for early 2026. The proposed system demonstrates that cost-effective, sensor- driven embedded technology can address a real-world road safety problem that continues to claim lives and that a working solution need not be complicated to be effective.

ACKNOWLEDGMENT

The authors gratefully acknowledge the guidance of Dr. Z. I. Khan (Project Guide) and Dr. V. B. Gadicha (Head, Department of Computer Science & Engineering, P. R. Pote Patil College of Engineering & Management, Amravati) for their continuous support, technical supervision, and encour- agement throughout this project.

REFERENCES

1. R. Baria and S. Degadwala, Emergency Vehicle Detection in Urban Traffic: A Deep Learning Perspective, Int. J. Sci. Res. Sci. Technol., vol. 12, no. 3, pp. 403410, May 2025.
2. S. D. Vartak and A. Sharma, A Systematic Study of Various Techniques of Obstacle Detection and Traffic Sign Detection Used in Self Driving Car Application of IoT, Int. Res. J. Adv. Eng. Hub (IRJAEH), vol. 3, no. 3, pp. 755759, Mar. 2025.
3. A. Praveena et al., Enhanced CNN Based Approach for IoT Edge Enabled Smart Car Driving System for Improving Real Time Control and Navigation, Sci. Rep., 2025.
4. T. E. Achar, C. Rekha, and J. Shreyas, Smart Automated Highway Lighting System Using IoT: A Survey, Energy Inf., vol. 7, p. 76, 2024.
5. A. Sharma, S. Kumari, and N. Singh, A Comparative Study on IoT Based Traffic Management System, IJRASET, 2023.
6. V. Adewopo and N. Elsayed, Smart City Transportation: Deep Learning Ensemble Approach for Traffic Accident Detection, arXiv, Oct. 2023.
7. R. M. Silva et al., Vulnerable Road User Detection and Safety En- hancement: A Comprehensive Survey, arXiv, May 2024.
8. M. Evans et al., Vehicle-to-Everything (V2X) Communication: A Roadside Unit for Adaptive Intersection Control of Autonomous Electric Vehicles, arXiv, Sep. 2024.
9. S. Usmankhujaev et al., An Autonomous Framework for Real-Time Wrong-Way Driving Vehicle Detection from Closed-Circuit Televi- sions, Sustainability, MDPI, 2021.
10. P. N. K. Murthy and A. M. Rao, Wrong-Way Vehicle Detection System for Highways Using Image Processing and Embedded Systems, Int. J. Eng. Res. Technol., 2019.
11. S. H. Kim et al., Design and Implementation of a Smart Road Safety System with Automated Spike Barriers for Wrong-Way Prevention, Int.

    J. Smart Home, 2017.

12. A. Al-Fuqaha et al., Internet of Things: A Survey on Enabling Tech- nologies, Protocols, and Applications, IEEE Commun. Surveys Tuts., vol. 17, no. 4, pp. 23472376, 2015.
13. Sydney Coordinated Adaptive Traffic System (SCATS), Roads andMaritime Services, NSW, Australia. [Online]. Available: https://www. transport.nsw.gov.au