Autonomous Robotic System Based Environmental

Assessment and Dengue Hot-spot Identification

Sudarshan Sreeram Lokesh Shanmugam
Vidya Mandir Senior Secondary School Inov Robotics Pvt. Ltd.
Mylapore, Chennai – 600004, India Avadi, Chennai- 600054, India
sudarshan.sreeram@gmail.com info@inovrobotics.com

Abstract— Robotic systems have a tremendous influence on advanced to a point where humans cannot live without it. One of
real-world applications as well as the actions and decisions of the most important factors involved in this advancement of
humans. It is one of the key driving forces towards advancements technology is robotic systems design. For decades, humans have
in the field of technology. With an everlasting desire to master this, been able to solve problems and achieve milestones using
humans have made one major compromise: unmonitored robotic technologies. For example, sending humans to the moon
depletion of Earth’s invaluable resources and uncontrolled and setting up space stations, decreasing cost and time of
accumulation of waste. Furthermore, this so-called progress manufacturing process using automated robots, and connecting
towards a technologically advanced world has paved the way to people all over the world through the internet to maximize
create a complex environment that poses several risks to human
productivity and communication.
health. One such risk is caused due to the stagnation of
contaminated water which serves as a breeding ground for insects Although we have made our mark as the most advanced
such as mosquitoes and houseflies and causes a plethora of species on the Earth, we have proved to be the most ignorant
diseases. As a result, mosquito-borne diseases are on the rise in when it comes to caring for our environment. On our path to
countries such as India, Sri Lanka, Thailand, Cambodia, and master technology, we have made some major compromises:
Brazil. These diseases are detrimental to the health of the unmonitored depletion of Earth’s invaluable resources and
country’s population. Dengue, a mosquito-borne disease, is the uncontrolled accumulation of waste. These compromises have
fastest spreading and most critical. Therefore, the challenge at
created a ripple effect leading to major problems all over the
hand is to develop a robotic system which is efficient in locating
world such as global warming and climate change, the spread of
“dengue hot-spots” while being complementary to the
environment, i.e. fewer emissions, zero greenhouse gases. Current
diseases and viruses, and a decrease in resources necessary to
methods of dealing with this challenge require a lot of resources sustain life on Earth. Our desire to invent the “next big thing”
and do not yield expected results. Our design methodology, has made us forget that we live in a world with limited resources.
referred to as EMBED-X, offers a systematic approach that is This lack of concern for the sustenance of the natural world has
backed up by protocols which support intelligent and pro-active paved the way for creating a complex environment which poses
monitoring of the environment. The robotic system is to be semi- several risks to human health.
autonomous and remotely connected to a control station for
In this complex environment, the spread of unwanted
constant data analysis. The aim is to identify objects such as
diseases and viruses is strikingly high due to its potential to
coconut shells, rubber tires, and plastic containers that have the
potential for retaining stagnant water. To identify these objects, sustain the lives of disease-causing organisms. For example,
image analysis algorithms are applied on the images captured by stagnation of contaminated water serves as a potential breeding
the autonomous robot during the surveillance of a given area. Our ground for insects such as mosquitoes and houseflies which
image analysis algorithm has an accuracy of 66.7% and is cause a plethora of diseases. These breeding grounds, usually
continually being improved using the EMBED-X methodology. located in inaccessible regions, accumulate and stagnate water
There are limitations to this solution and methodology. However, over a period of time. In countries such as India, Cambodia,
this project would serve as a stepping stone towards better design Thailand, Brazil and Sri Lanka, the supporting nature of the
of robotic systems applied specifically to environmental tropical weather has led to the increase in a number of breeding
monitoring and improving assessment capabilities through grounds for these insects. As a result of this, mosquito-borne
advanced image processing methods. diseases are on a rise in these countries and cause detrimental
health complications for the affected victims. One of the most
Keywords—EMBED-X; Systematic Approach; Image Analysis widespread mosquito-borne disease that is ranked “critical” by
Algorithms; Design Methodology; Mosquito-borne diseases; the World Health Organization (WHO) is Dengue [1]. In the
year of 2017, India faced 157,220 cases and 250 deaths caused
I. INTRODUCTION
due to the dengue virus – the highest ever recorded number of
Technology, as we know it, is on an exponential rise and cases and deaths over the past eight years [2]. For the present
serves as the fundamental basis for many real-world applications paper, the geography is based in Tamil Nadu, India.
such as industrial and commercial robotics, smart home
appliances, construction equipment and machinery, and Since the spread of dengue has proven to be an increasing
vehicular transport systems. This is only the tip of the iceberg concern in tropical as well as sub-tropical countries, the
that covers the many innovations of technology. Technology has challenge at hand is to develop and implement a robotic system

978-1-5386-5186-5/18/$31.00 ©2018 IEEE

that can locate these breeding grounds while being However, it is known that improper use of insecticides not only
complementary to our environment. Although these breeding cause harm to the environment, but also affects the health of
grounds support the spread of various diseases, they are more plants and animals alike [14]. A degree of commonality arises
commonly known as “dengue hot-spots”. There are, of course, between these papers in terms of agreeing with the fact that
many methods for solving this challenge: dengue vaccines drones have been proven to be effective in locating dengue hot-
(which are currently unavailable in India), mosquito repellent spots. Thus far, focus of previous research in this context have
products, and mosquito prevention traps. Some solutions even been reactive rather than pro-active. The development of a
resort to the burning of worn out tires, disposed plastic robotic system for pro-active monitoring of our environment
containers, and broken coconut shells which have the potential requires a carefully planned step-wise approach to maximize
to store and stagnate water. These solutions are deemed efficiency and to yield expected results.
ineffective due to the unavailability of vaccines and repellents
and contribution to existing environmental problems due to the To achieve this, our design methodology, referred to as
burning of environmental waste. For example, burning of EMBED-X, is used to systematically develop and deploy a semi-
environmental waste is nothing but a careless attempt to autonomous robotic system that can use an image analysis
eliminate the possibility of breeding grounds since this activity algorithm to help identify specific environmental wastes in
inaccessible regions. EMBED-X is based on the internationally
contributes to the already critical problem of global warming.
Furthermore, mosquito repellent products and mosquito acclaimed six sigma design methodology and offers a
prevention traps do not eliminate the source of the disease, but sustainable, yet effective solution to this challenge. This
rather proposes to reduce the probability of humans from being methodology is backed up by protocols which offer support for
affected by them. In addition, these products and solutions have intelligent and pro-active monitoring of the environment. It
little or no effect on the number of dengue cases every year in allows for the development process to evolve over a series of
India as shown in fig. 1 [3, 4, 14]. It is not surprising to see that steps: requirements collection, customer surveys, consideration
the number of cases only increase every year despite the efforts and evaluation of the requirements, design and implementation,
taken by environmental organizations as well as governmental and deployment and further development. As the complexity of
bodies. The trendline, represented in yellow, clearly shows an a system increases, it raises the importance of keeping the
increasing trend. One issue with these solutions is that they are development process clear, simple, and modular.
aimed at inducing the confidence that “you are safe” but fails to Following this approach, a robotic system was designed to
address the fact that our environment is degrading and being include synchronization between a quad blade drone and a
disrupted due to our own actions and decisions. compact rover. Both these robots are fitted with high definition
cameras to capture images for object identification. A Linux
based computer, connected to both the drone and the rover
wirelessly, is used to perform the image processing. This
computer is referred to as the control station. It is wirelessly
connected to both the drone and the rover for managing the semi-
autonomous process as well as processing information supplied
from the robots.
The dengue hot-spot identification process starts with the
quad blade drone surveying a given area and capturing images.
These images are processed using the image analysing algorithm
Scale Invariant Feature Transform (SIFT) [10] to identify
coconut shells, disposed of plastic containers, and worn out tires.
GPS coordinates of the object are then extracted from the
geotagged images and sent to the rover. The rover is then
Fig. 1. Graphical representation of dengue cases in Tamil Nadu, India directed to the specified coordinates and perform an
[Data from NVBDCP, India] environmental assessment of the surrounding area by capturing
high definition images. These images are then analysed by our
Apart from these solutions, several research papers [5,6,7] algorithm which has an accuracy of 66.7%. This is followed by
have come up with innovative ideas to tackle this challenge. A an analysis of the weather records and forecast for that area. This
research paper – from IIT Bombay, India – discussed the use of is done to check for rainfall. The rover then accordingly updates
an image analysing drone to search for stagnant water [8]. This a virtual map of all known dengue hot-spots with the required
paper expresses the difficulty in locating water in an image when information. With this data, several methods such as larval
using image analysis algorithms and provides a solution for the source management – although not a part of this paper – could
same. Although it is an interesting proposal, one of the be used to clear the dengue hot-spots so that pro-active
limitations in their proposed solution is that it is not necessary monitoring and prevention of dengue may be achieved.
for all active breeding grounds to be aerially visible. Some of
them may be hidden away – possibly under some debris or This paper is organized as follows: section 2 and 3 cover the
within a densely populated collection of plants. So, the drone theoretical and practical implementation, accompanied by the
searching specifically for water may miss a dengue hot-spot results while section 4 covers the analysis of results and
which is unacceptable. Several other papers use drones to limitations of this research, followed by main conclusions.
automate the process of the spraying of insecticides [9].
 II. EMBED-X METHODOLOGY SUMMARY module which allows it to communicate with the control station
and provide a live status of the assessment process. The drone,
EMBED-X is an acronym for the six-stage design process on the other hand, is provided with a high definition camera
that includes the following phases: Enumerate, Measurement, fitted on a 3-axis gimbal for stable image capturing. The drone
Bottlenecks, Evaluate, Design, and Phase X which is meant to also connects to the control station, through the same semi-
signify continuous improvement [15]. autonomous process used by the rover, to upload its status.
Enumerate: In this phase, the system design problem is Images from the drone, just like those from the rover, are
defined capturing the requirements of customers. These manually transferred to the control station. Therefore, the system
requirements are obtained through surveys and customer combines the use of data collected from both aerial and ground
interviews. monitoring to create a massive set of data that would be
uploaded to an online virtual map and used in the future to
Measurement: Analysis of the survey measures is carried out eliminate sources of the disease. An illustrative representation of
during this phase. The analysis includes the customer survey the process described above is depicted in fig. 2. The key
information, consolidation of design specification and requirement in this process is the inter-dependency between the
constraints. two robots since it is essential for the integrity of the system. The
Bottlenecks: This is an elaborate phase where several combined drone-rover system execution process takes place
development activities such as software or hardware through well-defined steps as shown in fig. 3.
developments are accomplished after the resolution of risks or
process bottlenecks.
Evaluate: The implementation of first-round testing of the
system is performed. It also includes the consolidation of a
system dashboard, configuring analytic engines (statistical
analysis) and establishing process control mechanisms.
Design: The final embedded system design is realized in the
stage after fine-tuning and trade-offs to the process parameters
in Phase 4. The trade-offs, if required, are affected by game
theoretic models (e.g. Nash or Kalai-Somordinsky solutions
[11]). This design should correspond fully to the requirements
stated in Phase 1. Fig. 2. A schematic view of a typical dengue infested environment
Phase X: This phase stands for continuous improvements or
Kaizen events.

III. ENVIRONMENTAL ASSESSMENT USING A ROBOTIC SYSTEM

Environmental assessment, especially in rough terrain
proves to be a complex challenge that needs to be approached
systematically. The proposed method aims to simplify that
challenge by tackling it step-wise using the EMBED-X
methodology. Using this, the autonomous robotic system was
designed specifically to meet the following requirements:
economically viable (low cost), easily upgradeable, structurally
durable, and remotely synchronizes with a computer for
measurements and data analysis.
The two custom-built robots, a powerful medium-sized
drone and a sturdy small-sized rover, form the basis of the
proposed robotic system. A computer (control station) based on
a Linux operating system is responsible for the management and
analysis of the data supplied by the system of robots. This
control station plays an integral role as it manages and maintains
the process flow.
Fig. 3. A diagrammatic representation of the combined drone-rover
As the robots are to be structurally durable and highly execution process
recyclable, the core structural components are made of strong
The constant data streams from the robots are transmitted to
and lightweight aluminum and carbon fibre. The compact rover
the control station through radio frequency technologies. Data
is fitted with a high definition camera that is used to capture
telemetry modules are attached to the robots as well as the
images. When the rover returns to the control station after its
control station. This allows for data transmission within a range
assessment, the captured images are transferred to the control
of 1km to 1.5km. Apart from this, both the robots are equipped
station via an SD card. The rover is fitted with a data telemetry
with flight controllers that can carry out pre-defined tasks. Since
both the robots are always connected to the control station, they
are provided with predefined mission paths. The module allows
for live monitoring of the robotic system during the execution
process. The data moves from the flight controller to the control
station through the data telemetry module.
A. Drone Specifications
The drone has a crash-resistant carbon fibre frame, an
advanced flight controller, and an accurate GPS. It is also fitted
with a 4000 mAh LiPo battery pack which offers a reasonable
period of flight time. Equipped with four 11-inch carbon fibre
blades, the drone can remain structurally stable even at high
altitudes (50 m to 75 m). A high definition camera is mounted
Fig. 5. Custom built rover without the mounted camera
on a 3-axis gimbal connected to the front of the drone. The drone
utilizes 9 kV brushless DC motors for rotating the four blades C. Image Analysis
and electronic speed controllers to control the speed of each
motor. As the drone is fitted with the data telemetry module, The high definition cameras used by the robotic system have
there is constant supply of data to the control station. Data from 12 mega pixel sensors and an image resolution of 4000 by 3000
the drone include altitude (measured using the barometer on the pixels. Both the cameras use a wide-angle field of view setting
microcontroller), GPS coordinates, speed, and direction when capturing the image. This helps in capturing a large area
(determined using the on-board compass). The mission planner easily while maintaining image detail. Both the drone and the
rover have an auto timer feature which allows the camera to
at the control station is used to create predefined flight plans for
capture images in 30 second intervals. The image analysis
the drone. The drone has a flight time of 20-25 minutes and
process is carried out at the control station which uses the ROS
includes protocols for when anything goes wrong during
package Find_Objects_2d [12]. This package takes advantage of
environmental assessment. As shown below, Fig.3 presents a
the object recognition algorithm SIFT to aid in object
view of the drone used in this study.
identification.
Training the algorithm is necessary to help identify different
type of objects considered in this study. It also helps in
increasing the algorithm’s accuracy and overall performance. To
train the algorithm, a set of ten images depicting the object
placed in front of a white background are used. The objects in
the image are marked by cropping them from the image
manually. These marked images are fed to the algorithm for
training. In the training process, the algorithm analyses the
cropped region and marks characteristic points (match points)
which may be unique to a given object. These match points are
stored under the name of that object. The algorithm compares
Fig. 4. Custom built quad blade drone the match points in a given image with those match points,
stored under the names of each object, to check for similarity.
B. Rover Specifications Fig.6 presents a case where none of the match points are similar
The compact rover has an aluminum frame for structural to those stored under the name of the objects. When the object
rigidity, four 300 rpm Johnson geared DC motors, a high is identified, it is outlined by a coloured rectangle as shown in
definition camera, and a 2200 mAh LiPo battery for long-range Fig.7. As the number of marked images fed to the algorithm
missions. The robot’s frame is fitted with ultrasonic sensors for increases, the number of characteristic points for each type of
obstacle detection and IR sensors for following a path. The object increases. This implies that the accuracy of the algorithm
ultrasonic sensor is mounted in the front along with the camera, in detecting an object increases with increase in number of
whereas the IR line following sensors are mounted below the marked images analysed.
rover. The rover is also fitted with the data telemetry module The target objects that our image analysis algorithm is
which allows it to communicate seamlessly with the control capable of identifying are rubber tyres, disposed plastic
station. Data from the rover include GPS coordinates, speed, containers, and coconut shells. Based on our survey and expert
direction, and sensor data (ultrasonic and IR sensors). The rover interviews with the swage control board, Chennai, India, the
has 4 thick rubber tyres which allows it to comfortably targets objects were identified for this study [14]. These target
manoeuvre rough terrains. The rover utilizes the same flight objects are considered based on a root cause analysis and support
controller used by the drone. This also allows the rover to carry the data related to the dengue issue that is currently being
out semi-autonomous processes. For example, the rover can experienced in Chennai, India.
follow a mission path and survey a given area on its own. Fig.4
represents a view of the sturdy rover without the mounted
camera.
 captured images were transferred via an SD card. After
successfully uploading the images, stage three was triggered. In
this stage, an image library consisting of images taken during the
rover’s assessment was created. Following this, the control
station was used to perform an analysis of the captured images
to confirm the presence of the target objects.

Fig. 6. Algorithm fails to identify plastic container

Fig. 8. An image taken by the drone when surveying the selected area
Fig. 7. Coconut shell is identified by the image analysis algorithm

IV. EXPERIMENTAL RESULTS The next stage was a very important since it decided the priority
The execution phase of the robotic system evaluated both the and probability level of the potential dengue hot-spot. For
hardware and the software to test for accuracy and performance. example, in fig.8, it is clear that there is presence of stagnant
For example, the algorithm for image analysis was repeatedly water and environmental waste. Hence, the location can be
tested and fine-tuned to maximize the accuracy of recognizing marked as a potential dengue hot-spot. Stage four required the
an object. This phase of the development process was critical as analysis of weather records and forecasts to check for rainfall. If
it allowed for the elimination of bugs and reduced the chance of the selected area had received rainfall in the previous three days,
possible errors in the final model. To conduct the experiment, an then it confirms a high probability of finding stagnant water. If
area within the city was selected based on the number of dengue not, then the robot has identified areas where water may stagnate
cases. in future. The last stage, stage 5, involved the creation of a
virtual map containing location pins of all confirmed and
The execution process was systematically carried out in five potential dengue hot-spots. Each location pin was accompanied
stages. Stage one involved the aerial photography of the selected by a data set that contained the priority and probability levels,
area using the custom-built drone. The selected area was marked weather forecasts and records, aerial and ground images, and
on a map in the mission planner by bounding it within a type and number of objects found.
rectangle. The drone covered the area within the rectangle by
following a path like that followed by a square wave. After
completing its assessment, the drone returned to the starting
position and landed automatically. The images from the camera TABLE I. TRIAL RUNS FOR OBJECT RECOGNITION
were transferred to the control station manually via an SD card
Tyre Plastic Container Coconut Shell
for image analysis. Fig. 8 shows one of the images captured by
the drone during the assessment process. Each image was Trial 1 Identified Not Identified Identified
analysed for the target objects and since the images were Trial 2 Identified Not Identified Identified
geotagged, the GPS coordinates of the objects could be extracted Trial 3 Not Identified Identified Identified
from the image. A mission planner was setup for the semi- Trial 4 Not Identified Identified Not Identified
autonomous rover and it was directed to the coordinates of the Trial 5 Identified Identified Identified
identified object. To perform this, the rover made use of the IR Trial 6 Identified Not Identified Identified
line following sensors to detect line paths drawn on the side of Trial 7 Not Identified Identified Not Identified
roads and move closer to the desired location. Along with this Trial 8 Identified Identified Identified
sensor, the ultrasonic sensor aided in detecting obstacles along
Trial 9 Not Identified Identified Identified
the line path. The robot reached a point till which the line path
was available and switched to a free-range mode a where it used Trial 10 Identified Identified Not Identified
only the ultrasonic sensor to reach the destination. Stage two
commenced when the rover reached the assigned location and Table I represents the results of our image analysis algorithm
captured images of the artificial surroundings for a closer view. on identifying three different objects. Our algorithm achieved an
The rover was directed back to the control station where the accuracy of 66.7% in finding the target objects. Table II
represents the object identification results which helped in A systematic approach helps in keeping the development
determining the accuracy of the algorithm. process clear, simple, and modular. The robotic system is semi-
autonomous and connected to a control station for constant data
TABLE II. OBJECT IDENTIFICATION RESULTS monitoring and analysis. Target objects such as coconut shells,
Object Successful Trials Failed Trials disposed of plastic containers, and rubber tyres are considered
Rubber Tyre 6 4 because they support the data corresponding to the dengue
Plastic Container 7 3 situation currently being faced in Chennai, India. A virtual map
of all identified and potential dengue hot-spots is created and
Coconut Shell 7 3
updated frequently. This massive pool of data would be used in
Total 20 10
clearing these hot-spots in the future. Although the proposed
solution has limitations, the goal of this paper is to serve as
There are limitations to the approach proposed here. For stepping stone towards a better design of robotic systems aimed
example, the flight time of the drone is restricted to 25 minutes at enhanced environmental monitoring capabilities by
which does not allow for region-wide analysis. Image analysis employing advanced image processing solutions.
is restricted to 2D objects with the current ROS package. The
lack of depth as a factor implies that 2D images would also be ACKNOWLEDGMENT
recognized as an object which is not desirable. Another issue
would arise when the line path for the rover is blocked or The first author would like to thank his teachers at Vidya Mandir
removed. This would halt the rover’s mission for an indefinite Sr. Sec. School, Chennai, India for their advice and support.
period of time leading to the rover getting lost. In addition to the
REFERENCES
above, scalability of the model with varied backgrounds would
require further testing and validation. [1] "Dengue | World Mosquito Program", Eliminatedengue.com, 2018.
Available: http://www.eliminatedengue.com/ our-research/ dengue-fever.
Following the EMBED-X methodology, the proposed [2] "NVBDCP | National Vector Borne Disease Control Programme",
solution has reached phase X which is meant to signify Nvbdcp.gov.in, 2018. http://nvbdcp.gov.in/den-cd.html.
continuous improvement. To refine this solution further, [3] "Mosquitoes are becoming immune to the popular insect repellant Deet,
development is to be carried out. Future development and experts warn", Mail Online, 2018. http://www.dailymail.co.uk/
health/article-2282969/Mosquitoes-immune-popular-insect-repellant-
upgrades to this solution include automating the image transfer Deet-experts-warn.html.
and virtual map updating process, increasing the drone’s flight
[4] C. Pennetier, V. Corbel, and J. Hougard, "Combination Of A Non-
time and capability to avoid collisions, utilizing LiDAR Pyrethroid Insecticide And A Repellent: A New Approach For
technology [13] in the rover, anti-theft measures and failsafe Controlling Knockdown-Resistant Mosquitoes" Published in The
protocols, and live 3D object recognition to identify an even American Journal of Tropical Medicine and Hygiene, Vol. 72, Issue 6,
greater number of target objects including highly deformed Jun 2005, p. 739 – 744.
objects. A long-term goal is to create a swarm of drones and [5] C. Suduwella, et al (2017), “Identifying Mosquito Breeding Sites via
rovers which can collectively perform tasks and communicate Drone Images”, Published in DroNet ‘17 Proceedings of the 3rd
Workshop on Micro Aerial Vehicle Networks, Systems, and
with each other. Applications, New York, USA.
[6] MC Elemmi et al., "Computer Vision Based Identification of Dengue
V. CONCLUSIONS Mosquitoes From Images", International Journal On Imaging Science
And Pattern recognition, 2017.
Over the last two decades, there has been an exponential [7] A. Amarasinghe et al., "A Machine Learning Approach for Identifying
growth in technological advancements. This growth is Mosquito Breeding Sites via Drone Images", published in the 15th ACM
impressive, but it has not taken place without compromises. Conference on Embedded Networked Sensor Systems (SenSys 2017).
Humans have risked not only their lives but also the lives of [8] M.G. Prasad, A. Chakraborty, R. Chalasani and S. Chandran (2015),
“Quadcopter based stagnant water identification” Published in Fifth
other living beings in the name of progressing towards National Conference on Computer vision, Pattern Recognition, Image
technological advancement. The waste generated due to this is Processing and Graphics), Pune, India.
generally ignored and not treated with importance. The [9] J. T. Amenyo et al., "MedizDroids Project: Ultra- low cost, low-altitude,
importance of clearing this waste is only understood when its affordable and sustainable UAV multicopter drones for mosquito vector
control in malaria disease management," IEEE Global Humanitarian
accumulation starts to trigger diseases and discomfort for Technology Conference, San Jose, CA, 2014, pp. 590-596.
humans. Dengue, a mosquito-borne disease, has been on the rise [10] "Scale-invariant feature transform", En.wikipedia.org, 2018.
in tropical and sub-tropical countries. The actions and decisions https://en.wikipedia.org/wiki/Scale-invariant_feature_transform.
of humans have led to the development of its sources, known as [11] T. Lensberg, "Stability and the Nash solution", Journal of Economic
dengue hot-spots. Current methods of managing these hot-spots Theory, vol. 45, no. 2, pp. 330-341, 1988.
haven proven to be inefficient despite the efforts taken by [12] "find_object_2d - ROS Wiki", Wiki.ros.org, 2018. [Online]. Available:
environmental organizations and governmental bodies. A http://wiki.ros.org/find_object_2d.
robotic system consisting of a drone and a rover is designed to [13] "Lidar", En.wikipedia.org, 2018. [Online]. Available:
https://en.wikipedia.org/wiki/Lidar.
locate these dengue hot-spots. This system is designed with the
[14] “Private Communication”, 2017, CMWSSB, Sewage Board, Chennai,
EMBED-X methodology as its backbone. India.
[15] L. Shanmugam and S. Sreeram (2018), “EMBED-X: An integrated
methodology for human - robotic systems development and interaction”
Under review for publication in an IEEE Conference.