ARTIFICIAL INTELLIGENCE ASSIGNMENT

IMPLEMENTATION OF AI IN TRANSPORT SECTOR

Presented By:
AROHI SHARMA(CMBA13)
PRAKHAR MEHROTRA(CMBA13)
SOHAIL PARVEZ(CMBA13)
SAPTASHWA MUKHERJEE(CMBA2Y2)
INTRODUCTION

Artificial intelligence has tremendous potential to contribute to global economic activity, but
to optimize the benefits, it will be important to navigate the widening disparities between
nations, companies and working men and women. (Mackworth, 2006)

The role of tools and techniques for artificial intelligence (AI) in business and the global
economy is a hot topic. This is not surprising given that AI will bring about radical; arguably
unprecedented changes in people’s way of living and working. The AI revolution is not in its
infancy, but it still must have most of its socio-economic impact. (H.M, 2000)

AI technologies may lead to a performance gap between leading marketers (companies that
will completely absorb AI technology across businesses for the five to seven years to come)
and non-adopters (companies that has little or no AI technologies fully absorbed by 2030).

The AI’s prospect is huge, yet its impact is undeniably disruptive. AI’s productivity dividend
is probably not immediately realized. The effect is likely to increase over time at an
accelerated pace; therefore, the short-term benefits of the initial investment may not be
evident. Long-term strategic thinking and persistence will be required.

It must be clarified that, even though we see economic potential in the use of AI
technologies, AI usage must always address concerns such as data protection, confidentiality
and possible bias issues. (YASDI, 2006)
The importance of AI, thus, is not found solely in the models themselves, but in the utmost
ability of organizations to leverage them.

AI (IN TRANSPORTATION INDUSTRY)

Artificial intelligence (by converting traffic sensors into cameras) is used in the prevention
and detection of traffic accidents and conditions, in the resolution of control and optimisation
problems, autonomous trucks have been launched around the world recently and are aimed at
reducing costs, reducing emissions and improving road safety compared to conventional road
transport systems.
1|AI
AI, in the fields of Transport and Infrastructure, can collect traffic data to reduce congestion
and improve public transport scheduling, Transportation is affected by traffic flow, AI
enables streamlined traffic patterns, Smarter algorithms and real-time tracking can efficiently
control high and low traffic patterns for optimal scheduling and routine. (H.M, 2000)

Artificial intelligence (AI) can be used for critical mission tasks, such as self-driving
passenger vehicles, to detect pedestrian and cycling trails, reduce traffic accidents and
injuries, make it possible to use transport more diverse and reduce emissions overall.
Machine-learning will soon be used to predict and avoid traffic jams, Artificial Intelligence
improves public safety, and the safety of people while commuting through public
transportation in urban areas which is further improved by monitoring crime data in real time.
(Mackworth, 2006)

The problems arise in transportation if the behaviour of the system is too complex to model
by way of a predictable pattern, affected by traffic, human errors or accidents. In those cases,
the unpredictability can be worked to help by AI, and the data observed can be utilized
thereby to make or predict decisions appropriately.

AI is driving up the evolution of next-generation intelligent transport systems, with its

burgeoning branch machine learning, making it possible for transport agencies, towns and
private car owners to take advantage of modern computer technologies and communication
technologies, making mobility much safer and robust. (Feigenbaum, 1982)
The possibilities for multiple application of big data and artificial intelligence in the transport
sector were created with high quality commercial CPU, GPU and IoT communication
technologies such as the LTE, 5 G and LPWAN.

Artificial intelligence can be used for traffic management and decision-making systems,
which improve and streamline the management of traffic and make our roads more
intelligent. Traffic management systems can recognize conditions physically and
environmentally which may give rise to heavy flows and congestion of traffic. This is being
dealt via suggesting further clearance routes on its own.

2|AI
There is now an all-inclusive app, which provides AI-powered, real time traffic alerts from
services including Google Maps. Such apps will soon be directly contesting autonomous
vehicle by using location data obtained from smartphones users, so they can forecast and
analyse traffic conditions in your area to notify your travel plans.

Thus, it can rightly be claimed that AI in transportation industry/sector contributes to

improving safety for passengers, reducing traffic congestion and accidents, reducing carbon
emissions and minimizing overall financial spending.

Artificial intelligence in road transport

Road transport is one of the most successful sectors in which AI has been applied, opening
completely new levels of cooperation between different road users. Car manufacturers,
technology firms and research groups around the world are pursuing AI technologies to build
automated vehicles for commercial as well as personal transportation use. Such vehicles are
based on a variety of sensors (such as GPS, cameras, radar) combined with actuators (devices
converting an input signal into motion), control units, and software. Some of these
technologies take over only certain driving functions (such as parking), others are meant to
replace the human driver completely. AI systems that take over certain driving functions are
already widely available on the EU market, while in a limited number of driving situations
and regions, fully automated vehicles are being tested (including transporting parcels).
(Burkov)

Testing automated vehicles in urban areas is typically more difficult, as there are many
different actors, complicated road systems, and infrastructure (intersections, road signs, etc.),
where the vehicle needs to anticipate much more (often unpredictable) signs of movement.

AI also makes it possible to platoon trucks–coupling several heavy goods vehicles (HGVs)
within a minimum distance from each other, allowing them to accelerate or brake
automatically and simultaneously. While a human driver drives the lead HGV, drivers in
those that follow may only be present in cases where complicated traffic conditions (such as
roundabouts) or unforeseen accidents occur, rather than driving actively. In the future, drivers
' roles of following HGVs are expected to decrease slowly until they are no longer needed in
the end.
3|AI
Although truck platooning is already being tested by HGV manufacturers in several EU
countries, in more complex traffic situations, further tests on multi-brand truck platooning are
still needed to verify that safety requirements are met.
However, AI algorithms are commonly used in the sharing of economic networks providing
road transport services. For example, Uber uses AI techniques in all aspects of its services,
from matching riders and drivers to optimizing the route.

In road traffic management, AI technologies are also used to help analyse traffic patterns,
volumes and other factors. In addition, these can provide drivers with data on the fastest route
to ease any traffic congestion that may have created. AI technologies also help keep traffic
flowing through traffic signals and traffic lights that rotate to meet on - the-ground traffic
flow requirements in real time.

Source: Autonomous vehicles and their impact on road transportation, 2018.

4|AI
Opportunities and challenges
AI brings great advantages to road transport but also presents serious challenges, especially
in mixed-use environments. AI has the potential to increase traffic efficiency, promote traffic
congestion, free driver time, facilitate parking, and enable car-and ridesharing. When AI
helps to keep road traffic flowing, vehicles idling can also reduce fuel consumption and
improve air quality and urban planning as well. Lower transportation costs and freeing the
driver from driving duties, however, could also result in more people choosing a vehicle as a
mode of transportation (instead of public transportation) and then increasing traffic and air
pollution. (Alpaydin)

Furthermore, cybersecurity and data privacy are also particularly important in automated
vehicle AI development. In other words, automated AI-based vehicles require access to a lot
of often sensitive or protected data. If third parties achieve uncontrolled access to automated
vehicle data, the safety of the vehicle, its occupants and other road users will be jeopardized.
AI-based systems used in autonomous transport could significantly improve road safety as
human error (such as speed, distraction, and consumption of alcohol) is involved in more
than 90% of accidents on EU roads; in which more than 25,000 people lost their lives in
2017.

Nonetheless, AI also creates new risky situations, as shown by accidents involving automated
vehicles. Drivers could be distracted and pay less attention to the road in an interim period
when vehicles are increasingly automated but not yet fully autonomous. A distracted driver
may be slow to react when a situation arises where the human needs to interfere. Maintaining
active visual attention during a longer period of automated driving is clearly difficult for
humans. (Norvig)

Therefore, there is currently a need for more progress to ensure that fully automated vehicles
can interact safely with other road users, perform well under all weather and road conditions,
recognize obstacles correctly and understand the environment.

The responses given to these issues by public and private investors will also assess public
acceptance and the consequent market penetration of road transport AI technologies. A recent
5|AI
Eurobarometer survey on autonomous systems found that respondents are more comfortable
with autonomous vehicles transporting goods than they themselves are riding in such a
vehicle. The study also showed that autonomous vehicles felt more comfortable with younger
respondents and respondents with a higher level of education. (Massaron)

Robots and AI in Logistics:

Global providers such as Amazon are taking great care to ensure high-level service delivery
and first-class customer experience. Logistics and supply chain business owners pay a lot of
attention to back-office automation in order to satisfy these goals. They delegate to robots
and then to an artificial mind repetitive and detail-oriented tasks.
Machine learning allows AI to accumulate information based on human decisions, to make
judgments and to interact with people. With the aid of robotic process automation, some
fewer complex tasks can be completed.
Based on rules, RPA-powered systems can help automate mundane tasks. Access to shipping
information, fill in some invoices and web forms, and transmit information to the right server
are among these activities. Eventually, AI will be able to access the data.

The use of robotic units for data-related repetitive tasks offer businesses significant perks.
Amazon employs more than 100,000 warehouse robots that can already take over from
human employees in many aspects. And as the statistics by EY – one of the largest
accounting firms – claims, this can enable costs saving up to 65%.

Technical Aspects of AI in Transportation:

6|AI
Currently, AI methods can be divided into two broad categories: (a) symbolic AI, focusing
on knowledge-based systems (KBS) development; and (b) computational intelligence,
including methods such as neural networks (NN), fuzzy systems (FS), and evolutionary
computing. A very brief introduction to these AI methods is given below, and each method is
discussed in more detail in the different sections of this circular.
Knowledge-Based Systems
A KBS can be defined as a computer system capable of providing advice in a domain, using a
human expert's knowledge. A distinctive feature of KBS lies in the knowledge separation that
can be interpreted in a few ways, such as rules, frames, or events, and the inference engine or
algorithm which uses the knowledge base to reach a conclusion.
Neural Networks
NNs are biologically based structures consisting of a computational "neuron" network,
arranged in layers. By changing the network weights, NNs can be "used" to approximate to a
desired degree of accuracy virtually any nonlinear function. NNs typically are provided with
a set of input and output exemplars. A learning algorithm (such as back propagation) would
then be used to adjust the weights in the network so that the network would give the desired
output, in a type of learning commonly called supervised learning.
Fuzzy Systems
Zadeh (1965) suggested Fuzzy set theory to address the uncertainty associated with almost all
real-world issues. Fuzzy set membership features provide a way to demonstrate that an entity
may be part of a group. Classic set theory establishes strong distinctions between sets,
meaning an object can only be a member of a given set or a non-member. Fuzzy membership
features enable gradual transitions for objects within sets between sets and varying degrees of
membership. A value of + 1 is indicated for complete membership in a fuzzy function, while
a value of 0 is indicated for complete non-membership. Partial membership is a quality
ranging from 0 to + 1.
An example of a fuzzy membership function defined on a certain highway for a collection of
"medium traffic volumes." Traffic volumes between 800 and 1,000 vehicles per hour (vph)
are completely part of the medium traffic level set in this example. Traffic volumes of less
than 400 vph or more than 1,400 vph would not be considered at all as a medium (member
feature value= 0). However, values between 400 and 800 vph, or between 10,00 and 1,400

7|AI
vph would have partial membership in the medium traffic level set. In a crisp set definition,
on the other hand, only values between 800 and 1,000 vph would be regarded as medium,
while all other values would not (for example, a traffic volume of 799 vph would not be
regarded as a medium traffic level). The use of fuzzy set theory does not necessarily
minimize uncertainty related to problem objectives or input values, but rather provides a
standardized way to systematically capture and define ambiguity. (Rashid, Make Your
Own Neural Network)

8|AI
 Language Use In AI:
● Python
● C++
● Java
Java, Python, Lisp, Prolong, and C++ are major AI programming language used for artificial
intelligence capable of satisfying different needs in the development and designing of
different software. It is up to a developer to choose which of the AI languages will gratify the
desired functionality and features of the application requirements.

● Python:
Python (official website) is among developers favourites programming
languages in AI development because of its syntax simplicity and
versatility. Python is very encouraging for machine learning for
developers as it is less complex as compared to C++ and Java. It
also a very portable language as it is used on platforms including
Linux, Windows, Mac OS, and UNIX. It is also likable from its features such as
Interactive, interpreted, modular, dynamic, portable and high level which make it
more unique than Java.

Advantages
● Python has a rich and extensive variety of library and tools.
● Supports algorithm testing without having to implement them.
● Python supporting object-oriented design increases a programmer’s productivity.
● Compared to Java and C++, Python is faster in development.

● C++:
C++ is the fastest computer language; its speed is appreciated for AI programming
projects that are time sensitive. It provides faster execution and has less response time
9|AI
 which is applied in search engines and development of computer games. In addition,
C++ allows extensive use of algorithms and is efficient in using statistical AI
techniques. Another important factor is that C++ supports re-use of programs in
development due to inheritance and data-hiding thus efficient in time and cost saving.

Advantages
● Good for finding solutions for complex AI problems.
● Rich in library functions and programming tools collection.
● C++ is a multi-paradigm programming that supports object-oriented principles thus
useful in achieving organized data.
● Java:
Java (official website) is another programming language to answer, ‘which
computer language is used for artificial intelligence?’ Java is also a
multi-paradigm language that follows object-oriented principles and
the principle of Once Written Read/Run Anywhere (WORA). It is
an AI programming language that can run on any platform that supports it without the
need for recompilation.
Java is one of the most commonly used and not just in AI development. It derives a major
part of its syntax from C and C++ in addition to its lesser tools that them. Java is not
only appropriate for NLP and search algorithms but also for neural networks.
Advantages
● Very portable; it is easy to implement on different platforms because of Virtual
Machine Technology.
● Unlike C++, Java is simple to use and even debug.
● Has an automatic memory manager which eases the work of the developer.
(Warwick)

Conclusion:
Artificial intelligence is generally seen as a supporting tool rather than serving as a
replacement for human intelligence and ingenuity. Although artificial intelligence is currently
struggling to complete common sense tasks in the real world, it is capable of processing and

10 | A I
analysing data troves much quicker than a human brain could. Artificial intelligence software
can then return and present them to the human user with synthesized courses of action. In this
way, humans can use artificial intelligence to figure out possible effects of each action and
streamline the decision-making process.
AI is beginning to make possible what was once thought impossible, like driverless cars, said
Russell Glenister, CEO and Curation Zone founder. Because of access to training data and
fast GPUs, which are both key enablers, driverless cars are only reality. A huge amount of
accurate data is required to train driverless cars, and speed is essential to the practice.
Transportation problems arise when system behaviour is too difficult to model according to
a predictable pattern, affected by things like traffic, human errors, or accidents. In such cases,
the unpredictability can be aided by AI.
Citizens ' security while traveling by public transport in urban areas is improved by real-time
tracking of crime data. This will also allow police to increase their effectiveness by patrolling
and safeguarding their citizens.
Self-driven cars and trucks have been of high interest in the last several years. In the
commercial sector, Uber and Elon Musk have produced self-driving trucks to reduce the
number of accidents on highways and increase productivity.
The impact of AI in transport:
In October 2016, Uber announced a driverless truck made by Otto that successfully drove
120 miles at 55 mph without any issues. Additionally, Daimler trucks has produced an 18-
wheeler semi-autonomous truck with an auto-pilot system.
Costs of labour in this sector will continually decrease with increased use of AI, providing
higher profits for industry players. The issue of long driving hours and stopping for a break
will no longer be a concern with fully automated fleets.
Beyond straightforward labour costs, safety and traffic accidents will be majorly affected by
AI. The number of accidents involving truck drivers at night is a large issue and can be
significantly improved with the use of smart unmanned vehicles. The personnel and financial
costs of these accidents are quite substantial. Auto-pilot or complete unmanned vehicles can
allow the driver to have a snooze without causing severe accidents. Some AI trucks even
have a special feature of predicting accidents as well as health issues of people around the
truck like detecting a heart attack and alerting the services automatically with the location
and details of diagnosis. (John D. Kelleher)

11 | A I
Bibliography
Alpaydin, E. (n.d.). Machine Learning: The New AI.
Burkov, A. (n.d.). The Hundred-Page Machine Learning Book.
Feigenbaum, E. (1982).
H.M. (2000).
H.M. (2000).
John D. Kelleher, B. M. (n.d.). Fundamentals of Machine Learning for Predictive Data
Analytics: Algorithms, Worked Examples, and Case Studies.
Mackworth, A. (2006).
Mackworth, A. (2006).
Massaron, J. P. (n.d.). Machine Learning for Dummies.
Norvig, S. R. (n.d.). Artificial Intelligence – A Modern Approach (3rd Edition).
Rashid, T. (n.d.). Make Your Own Neural Network.
Rashid, T. (n.d.). Make Your Own Neural Network.
Warwick, K. (n.d.). Artificial Intelligence: The Basics.
YASDI. (2006).

12 | A I