Enhancing Animal Safety Measures by Avoiding Train Collisions

Using Internet of Things (IoT) with Advanced Sensors Association

S. Praveen Kumar, S. Kaviraj, R. Naveen Kumar and T. Nagaarjunan

Department of Computer Science and Engineering, Mahendra Engineering College, Tamil Nadu, India

Keywords: Animal Safety, Train Collision, Internet of Things, Accident Avoidance, IoT, Sensor, Train Accidents,

Wildlife Safety.

Abstract: Transportation infrastructure degrades ecosystems in part because animal deaths caused by car accidents

disrupt ecosystem dynamics and endanger vulnerable species. Despite the prevalence of wildlife train crashes

globally, this issue has received comparatively less attention in the realm of railroads compared to highways,

where it has been well studied and mitigated. The Naxalites in India will likely try to derail our protest train

by releasing tracks, and the Indian railways might be in danger from an animal-rail collision in a forested

region. Currently, behind China, Russia, and the US, India's railway network management ranks fourth

globally. India is a good case in point; in the last five years, the Indian government has enacted a plethora of

legislation meant to safeguard wildlife sanctuaries and jungle creatures living in close proximity to railway

tracks. Our analysis of crack detection and animal-rail collisions in the surrounding region will help us figure

out how to fix this issue and whether or not these problems might drive up the budget for the Indian Railways

and cause casualties and damage to property. In this paper, we present a cost-effective solution to the problem

of managing rail-animal accidents. We propose using load cells to receive the precise location of the defective

area in the track, which will help with both the protection of animals and other living beings from being run

over by trains and the monitoring of the track's integrity. Many lives can be spared if the affected area is

promptly corrected.

1 INTRODUCTION

Complex interactions between transportation

networks and wild animals are well-documented

(Seiler A, et al., 2017). Numerous species see a

decline in population near roadways, which has the

ability to change community make-up and ecological

processes due to habitat loss, fragmentation,

degradation, and direct death. Some species thrive in

the areas immediately next to the roadway, while

other species are attracted to those areas despite the

high mortality rate, suggesting that roadways do not

carry a universally negative impact on wildlife.

Because railroads are less ubiquitous than highways,

or possibly because they offer lower risk to humans,

strikes on trains have attracted less attention than

strikes on roadways. However, there is evidence that

train strikes can have an impact on populations, and

animals are occasionally hit more frequently by trains

than by roads nearby. Rail workers are much more

motivated to reduce strikes if the populations they

serve are vulnerable or endangered, or if the species

they represent is charismatic, keystone, or socially

significant (N.Selvakumar, et al., 2020),( Irene

Nandutu, et al., 2022),( Surbhi Gupta, et al., 2021).

On railroads, the most effective strategies for

lowering the number of wildlife-vehicle accidents

never work. In order to decrease the incidence of

wildlife-vehicle accidents by as much as 80% while

still allowing ecosystem connection, wildlife

exclusion fences and crossing structures are being

used more and more. Although animals killed by

automobiles have consumptive, passive-use, and

management values, these road mitigation measures

are expensive. There may be a better use of resources

on trains, since strikes seldom harm humans. In areas

where rail traffic is much lighter than on regular

roads, it may not be essential to exclude animals from

advantageous foraging, travelling, and living options

along trains by erecting exclusion fence. Alternatives

to exclusion fence include warning signs or animal

detection systems that increase driver awareness,

which can help them slow down in the event that they

see an animal, thereby avoiding a collision with

680

Kumar, S. P., Kaviraj, S., Kumar, R. N. and Nagaarjunan, T.

Enhancing Animal Safety Measures by Avoiding Train Collisions Using Internet of Things (IoT) with Advanced Sensors Association.

DOI: 10.5220/0013871300004919

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies (ICRDICCT‘25 2025) - Volume 1, pages

680-687

ISBN: 978-989-758-777-1

wildlife. The opposite is true for train operators, who

need minutes of notice time to safely slow down and

cannot alter their path. Road and rail accidents

involving animals are likely lessened by reducing

speeds systematically, which leads to shorter stopping

distances. Animals may run off of the trail if they are

fleeing from thick snow, steep terrain, or bodies of

water; these actions nullify the effect of reduced

speed unless it is reduced severely (Gayathri K K, et

al., 2023), (S.R. Mathu sudhanan, et al., 2025),

(Shanaka Gunasekara, et al.,2021).

Another way to lessen the likelihood of wildlife-

train incidents is to make it more likely that animals

will get off the track when they see a train coming.

Animals and humans alike are at risk of direct

collision or an abnormal flight response, maybe

brought on by fear, if they are unable to recognize an

approaching train. Elements such as vegetation,

terrain, heavy snow, particularly along track

deviations, or competing stimuli from neighboring

rivers or roads may have a large impact on the

visibility and audibility of an approaching train and,

thus, increase the likelihood of these detection

failures occurring. It is reasonable to presume

increased accident risk exists whenever these factors

are present at locations that the animals actively

frequent. Certain scenarios may benefit from the

given a warning signal, one that could not be covered

or disguised, in advance of train arrival to lessen the

likelihood of detection failures. If the signals were

delivered at a steady success time clocked relative to

the train's arrival, and whether stimuli providing

warning signals or conditions related to train arrival

are different, animals could learn to associate warning

stimuli in advance of train arrival (Shizhong Zhao, et

al.,2024). The signal does not have to be aversive, as

ignition near the conveyance in motion is an aversive

unconditioned stimulus. A wildlife caution system

recently utilized these behavioral principles in similar

fashion to human road-railway crossing lights. These

systems are effective, but they are costly and

proprietary, and they need tight interaction with

railway infrastructure. Roadside wildlife warning

systems, including deer whistles and headlight

reflectors, that are less expensive, are mostly useless.

The reason behind this might be because when a

vehicle is close by, reflectors and whistles fail to offer

the same level of spatial and temporal accuracy as

conditioned warning cues (Julia Milewicz, et al.,

2021), (ISTIAK MAHMUD, et al.,2023). We detail

an electronic system that integrates the active warning

system's pinpoint signaling with the passive warning

system's adaptability to different installations and low

cost in order to lessen the frequency of wildlife-train

accidents.

2 RELATED WORKS

There has been a steady rise in the number of vehicle

accidents resulting in injuries and fatalities to both

humans and animals throughout the globe (Atri

Saxena, et al.,2020). Therefore, AVCs, which include

wildlife species, pose a serious risk to road safety. For

road safety and wildlife conservation, it is necessary

to implement a mitigation strategy that will decrease

the frequency of vehicle-wildlife accidents. An

innovative approach for detecting animals and

avoiding collisions utilizing object detection is

presented in this research. For animal detection, the

suggested approach considers neural network

architectures such as SSD and the quicker R-CNN.

This study creates a new dataset with 31,774 photos

from 25 different animal groups. After that, a model

for animal detection is created using SSD and quicker

R-CNN object detection. Mean average precision

(mAP) and detection speed are the metrics used to

compare the new method's performance to that of the

current one. Using faster R-CNN and an SSD, the

suggested technique achieves a detection speed of

82.11% at 10 fps and a mAP of 80.5% at 100 fps.

Wildlife populations are further threatened from

tragic animal incidents on railroads (Balaji Kannan, et

al.,2024). Official figures suggest, on average, twenty

elephants a year die from train accidents across the

nation. Likewise, railroad organizations are

conducting the research required to find the causes and

prevention methods of these occurrences. Regardless,

many incidents are occurring on the railroad lines,

most of them near woodland regions. A well-defined

technical system is required to warn animals to stay

away from the railroad lines in order to lessen the

number of wildlife fatalities caused by these hazards.

Using Machine Learning (ML) frameworks, this

research finds ways to alert both the animals and the

loco pilot as they approach railway rails.

Nearly 190 elephants have perished as a result of

train collisions in India during the previous 20 years,

according to a study by the Ministry of Environment,

Forest, and Climate Change (Kandikonda S V S K

Devi Prakash, et al.,2023). As noted by Indian

Express, from April 2019 to March 2020, trains

supposedly ran over considerable amounts of animals.

Trains operating below capacity killed 2700 cattle

from 2020-2021. According to multiple sources, the

loss of livestock has slowed down a large amount of

rail journeys. Using a railway line tracking system is

Enhancing Animal Safety Measures by Avoiding Train Collisions Using Internet of Things (IoT) with Advanced Sensors Association

681

proposed in this study as a means to lessen animal

mortality. Protecting animals from harm in the event

of a train crash is the driving force behind this study's

development of a system for object movement

detection. Ultrasonic and proximity sensors are used

to detect the object's motion. The impending train is

alerted to by the loco-pilot and the centralized system

by means of an ultrasonic sensor, which also activates

a bell. As a result, it keeps animals safe and helps them

survive. Rails are considered to be around 5 kilometers

from the accident site when the system is triggered. If

not, they do nothing.

Among the main cities in the world, railways are

the most popular and the most widely used mode of

transit (Madhupriya, et al.,2024). The proposed study

discusses self-reliant and IoT-based real-time

monitoring and control. The Internet of Things has the

ability to improve aspects of the railway system.

Automating railways can significantly reduce

accidents and bring the technology up to date for

antiquated legacy systems. We have tested and

improved the proposed methods of human and animal

identification. Automate the monitoring of many

railway-related metrics and provide real-time control.

Create an automated system that uses less human

effort while preserving energy.

Rail operations and passenger safety are

jeopardized in the event of a wildlife-train accident

(WTC), especially with big creatures. To assess the

most perilous WTC sites and their dispersion, we

investigated 1,909 WTCs that took place in the Czech

Republic from 2011-2019 (this study (Vojtěch

Nezval, et al.,2020)). The 208 WTC hotspots were

identified using the KDE technique. They represented

0.7% of the length of the Czech rail network, and

contained 782 accidents (41.2%). By using a

collective risk metric, we located and ranked the WTC

hotspots. More WTCs per unit area occurred near

forests or streams than in other locations on the Czech

rail network, and fewer along agricultural, urban and

industrial land use. Moreover, as most WTCs occurred

in less than 1% of the train network, these results could

inform placement of crash safety intervention.

3 METHODOLOGY

An approach to animal accident prevention that

makes use of load cells and the Internet of Things

(IoT) and Arduino UNO is the central component.

The train's speed will be reduced and it will stop

moving until an object in its path is identified by the

load cell microcontroller. When the sensor senses

that there is no obstacle in its route, the controller will

stop regulating the brakes. The data is instantly

visible on the screen and may also be sent to the

designated individual via the IoT Module.

(a) Power Supply: A transformer is used to reduce

the alternating current (ac) voltage from its

normal 220V rms level to the level of the

required direct current (dc) output. The full-

wave rectified voltage is first filtered using a

simple capacitor filter to create a direct current

voltage, and then it is supplied by a diode

rectifier. Ripple or ac voltage volatility is a

common feature of the resultant dc voltage.

Regulator circuits eliminate ripples and

maintain a constant direct current value

regardless of variations in the input voltage or

the load linked to the output voltage. One of

the common voltage regulator integrated

circuits (ICs) is typically used to provide this

voltage control.

(b) Transformer: From a voltage range of 0-230V,

the potential transformer will reduce it to a

range of 0-6V. The next step is to link the

potential transformer's secondary to the op-

amp-built precision rectifier. Using a

precision rectifier has several benefits, one of

which is that it produces DC output at the peak

voltage, while other circuits simply produce

RMS output.

that uses four diodes in series. The two corners

of the network that are diagonally opposite to

each other receive the input to the circuit,

while the other two corners provide the output.

(d) IC Voltage Regulators: A voltage regulator is

an example of a common integrated circuit.

The circuits which comprise a regulator

embody a source of reference and a

comparator amplifier in addition to being a

control device and an overload protector. IC

devices can regulate an adjustable voltage, a

constant negative voltage, or both. The

regulators have power ratings from milliwatt

to tens of watts, and load currents in the

hundreds of milliampere and tens of amperes

are compatible with these values. The

following figure 1 shows the block diagram of

the proposed system.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

682

Figure 1: Block diagram.

(i) Arudino UNO: One board that uses the

ATmega328 microcontroller is the Arduino Uno. It

features a 16 MHz crystal oscillator, six analogue

inputs, fourteen digital I/O pins (six of which may be

used as PWM outputs), a power connector, an ICSP

header, a reset button, and a USB connection. It

comes with all the necessary components to support

the microcontroller; all you need is a USB cable, an

AC-to-DC adapter, or a battery to begin. The

following figure 2 shows the Arduino UNO.

Figure 2: Arduino UNO.

(ii) NodeMCU ESP32 IoT Module: The

computational power and inbuilt WiFi and Bluetooth

connectivity of ESP32 NodeMCU are making them

increasingly popular for use in making linked

products. Thanks to its breadboard-compatible

architecture and ease of programming in the Arduino

IDE, the NodeMCU-ESP32 makes comfortable

prototyping a reality. With this board, you get a BT

wireless connection in addition to 2.4 GHz dual-mode

WiFi. The following figure 3 shows the NodeMCU

ESP32 IoT Module.

Figure 3: NodeMCU ESP32 IoT module.

(iii) IR Sensor: Electronic devices that generate light

in order to detect certain environmental factors are

known as infrared sensors. Infrared (IR) sensors can

detect motion and also monitor an object's

temperature. In contrast to active IR sensors, which

generate infrared light, passive IR sensors only detect

the presence of infrared light. Every item typically

emits some kind of heat radiation in the infrared

range. An infrared sensor may pick up on these forms

of radiation that aren't visible to the human eye. An

infrared light-emitting diode (LED) serves as the

source of light, while a photodiode, which is sensitive

to infrared light of the same wavelength, acts as the

detector. These output voltages and the resistances

fluctuate in direct proportion to the intensity of the

infrared light that hits the photodiode. The following

figure 4 shows the IR Sensor.

Figure 4: IR Sensor.

(iv) Buzzer Unit: Buzzers and beepers are electrical

signaling devices that are often seen in vehicles, home

appliances (such microwave ovens), and game shows.

Typically, it's made up of a number of switches or

sensors that are linked to a control unit. This unit then

determines if a certain button was pressed or if a

certain amount of time has passed. In most cases, it

lights up the corresponding button or control panel

Enhancing Animal Safety Measures by Avoiding Train Collisions Using Internet of Things (IoT) with Advanced Sensors Association

683

and makes a constantly beeping or intermittent

buzzing noise as a warning. An electromechanical

system, similar to an electric bell but lacking the

metal gong (the source of the ringing sound), was the

original basis of this apparatus. These devices would

frequently be fastened to the ceiling or wall and

would utilize them as a kind of soundproofing. The

first electromechanical buzzers, which were powered

by stepped-down AC line voltage at 50 or 60 cycles,

gave rise to the name "buzzer" due to the rasping

sound they produced. The usage of a ring or beep is

another typical method of signaling the pressing of a

button. The following figure 5 shows the buzzer unit.

Figure 5: Buzzer unit.

4 RESULTS AND DISCUSSIONS

Using deep learning neural networks, an IoT-based

surveillance system was created to identify animals

on railway lines. Cameras and sensors in this system

gather real-time data that is analyzed to find animals

on the rails. Alerts are given to train operators upon

detection, hence allowing quick intervention to stop

collisions. Another method uses image processing to

find hazards-especially animals-on train tracks using

an AI-powered early warning system. The technology

notifies trains via IoT apps and Bluetooth transmitters

and uses machine learning algorithms to find animals.

A buzzer notice warns the train driver to slow down

or halt upon detection, hence lowering the possibility

of crashes. Using the learning algorithm for object

recognition, an Artificial Intelligence (AI) I and IoT

based train collision avoidance system captures real-

time photos of railway tracks with cameras. The

technology emails notifications to authorities,

triggers a buzzer to notify the train driver, and shows

real-time updates on an LCD screen, so guaranteeing

prompt intervention to stop crashes upon finding an

obstruction, such as an animal. Using image

processing technology, a smart siren system finds

animals close to train tracks. Cameras record

photographs of the surrounding region; computer

vision algorithms then analyze them to find animals.

The technology activates a siren to warn animals of

an approaching train, therefore motivating them to

leave the tracks and so lowering the risk of accidents.

AI-based technologies have been used in India to

shield elephants from rail accidents. These gadgets

employ movement patterns and infrared imaging to

identify elephants using sensor technology. Alerts are

transmitted to railway and forest agencies as

elephants are discovered, so allowing train

conductors to halt or slow down and so avoid

accidents. Likewise, Norway has used IoT

technology to safeguard reindeer against rail

accidents. The technology alerts train operators when

they are nearing the reindeer by comparing geo-fence

regions with GPS data from reindeer collars, so

enabling them to take required measures. By use of

IoT and artificial intelligence technology, these

systems and implementations show how well they

may prevent train-animal collisions, hence improving

safety for railway operations as well as wildlife.

These systems and studies show how well combining

IoT and artificial intelligence technologies prevents

train-animal collisions, therefore improving safety for

railway operations as well as wildlife. The proposed

website design's welcome home page outcome is

depicted in the accompanying image, Figure 6.

Figure 6: Welcome message.

Figures 7 and 8 show the suggested hardware unit

design and the findings of the animal detection status

display clearly, illustrating the proposed technique.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

684

Figure 7: Animal detection.

Figure 8: Proposed hardware unit.

The results of the suggested method's crack

detection status display and the emergency alarm

message raised by IoT web interface are clearly

shown in Figures 9 and 10, respectively.

Figure 9: IoT emergency alert notification.

Figure 10: Crack detection.

Figure 11 shows the results of a cross-validation

between the suggested method, which makes use of

an ESP32 IoT enabled module, and the standard

design, which relies solely on an Arduino UNO

controller. Table 1 provides a descriptive

representation of the same.

Table 1: Analysis of Detection Accuracy Between

Proposed Esp32 Module and Normal Arduino Uno Based

Design.

S.No. Days Arduino UNO (%) ESP32 (%)

1. 5 87.26 97.94

2. 7 85.54 97.14

3. 10 84.39 97.41

4. 14 87.72 97.38

5. 15 89.29 98.54

6. 18 86.51 98.59

7. 27 84.73 98.47

8. 29 85.95 98.63

9. 33 86.18 97.52

10. 36 84.40 98.29

Enhancing Animal Safety Measures by Avoiding Train Collisions Using Internet of Things (IoT) with Advanced Sensors Association

685

Figure 11: Detection accuracy.

Figure 12 shows the data loss ratio comparison

between the suggested method ESP32 IoT enabled

module and the standard design, which relies solely

on an Arduino UNO controller. Table-2 provides a

descriptive representation of the same.

Table 2: Loss ratio comparison between proposed ESP32

module and normal Arduino UNO based design.

S.No. Days Arduino UNO ESP32

1. 5 8.54 1.39

2. 7 9.36 2.16

3. 10 11.52 2.54

4. 14 13.86 2.89

5. 15 15.26 3.24

6. 18 17.69 3.78

7. 27 19.14 3.99

8. 29 22.36 4.16

9. 33 26.34 4.53

10. 36 28.52 4.61

Figure 12: Loss ratio analysis.

5 CONCLUSIONS

To improve railway safety and protect wildlife, it may

be necessary to install warning systems that are

activated when trains encounter certain animals. Load

cells, internet of things (IoT) sensors, and automation

based on the Arduino platform work together to make

this system capable of detecting obstacles on railway

lines and reacting accordingly by notifying

authorities and reducing train speeds. By combining

infrared sensors with LCD monitors, we can monitor

in real-time and respond quickly to avoid mishaps.

Reducing the financial burden and animal casualties

caused by train-wildlife incidents, this novel solution

improves railway operations while also harmonizing

with environmental preservation initiatives. Train

travel will become safer and more environmentally

friendly as new technologies allow for additional

iterations of this system. Worldwide, transportation

networks pose a threat to both humans and animals.

This paper proposes a novel solution.

REFERENCES

Atri Saxena, et al., "An Animal Detection and Collision

Avoidance System Using Deep Learning", Lecture

Notes in Electrical Engineering, 2020.

Balaji Kannan, et al., "Artificial Intelligence based System

to Prevent Animal Accidents in the Railway Tracks",

8th International Conference on Inventive Systems and

Control, 2024.

Gayathri K K, et al., "Early Warning System to Prevent

Animal-Train Collision", International Journal of

Scientific Research in Science and Technology, 2023.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

686

Irene Nandutu, et al., "Intelligent Systems Using Sensors

and/or Machine Learning to Mitigate Wildlife–Vehicle

Collisions: A Review, Challenges, and New

Perspectives", Sensors, 2022.

ISTIAK MAHMUD, et al., "Advancing Wildlife

Protection: Mask R-CNN for Rail Track Identification

and Unwanted Object Detection", IEEE Access, 2023.

Julia Milewicz, et al., "Methods to reduce wildlife

collisions with rail vehicles", Rail Vehicles, 2021.

Kandikonda S V S K Devi Prakash, et al., "Arduino-based

Railway Line Tracking System for Mitigating Animal

Accidents", International Conference on Computing

Methodologies and Communication, 2023.

Madhupriya, et al., "Detection and Accident Prevention of

Animals in Railway Track using AI and IoT",

International Journal for Multidisciplinary Research,

2024.

N.Selvakumar, et al., "Wildlife and Train Collision

Detection System by Wsn", International Journal of

Advances in Engineering and Management, 2020.

S.R. Mathu sudhanan, et al., "Using Mask R-CNN and

distance transform to detect elephants on railway tracks

for collision prevention", Measurement, 2025.

Seiler A, et al., "Wildlife Deterrent Methods for Railways-

An Experimental Study", Springer, 2017.

Shanaka Gunasekara, et al., "A Convolutional Neural

Network Based Early Warning System to Prevent

Elephant-Train Collisions", IEEE 16th International

Conference on Industrial and Information Systems,

2021.

Shizhong Zhao, et al., "Passive Safety Assessment of

Railroad Trains in Moose Herd Collision Scenarios",

Sustainability, 2024.

Surbhi Gupta, et al., "Deep Vision Based Surveillance

System to Prevent Train-Elephant Collisions",

Research Square, 2021.

Vojtěch Nezval, et al., "Spatial analysis of wildlife-train

collisions on the Czech rail network", Applied

Geography, 2020.

Enhancing Animal Safety Measures by Avoiding Train Collisions Using Internet of Things (IoT) with Advanced Sensors Association

687