Applications of Artificial Intelligence in
Transportation Systems: An Overview
Dr. Mohamed Elshenawy
Assistant Professor, Zewail City of Science and Technology
2019-06-23
Overview
• Why we need AI?
• What are intelligent machine?
• How can AI change our future? Transportation as an example.
• Things to consider when implementing AI solutions.
2019-06-23
2
Why we need intelligent machines?
•Empower humans with technology and qmplify our human
intelligence
• Ability to analyze large volumes of data to provide a basis for various
human decisions.
• Transform the ways we do things and drive more value.
• Increased productivity and ability to automate repetitive tasks.
• Higher Efficiency.
2019-06-23
3
How humans assess different situations and make
decisions?
2019-06-23
4
Dr. Mica Endsley's model of SA.
How humans assess different situations and make
decisions?
2019-06-23
5
Perception
“Perceive the status,
attributes, dynamics of
relevant elements in the
environment”
Endsley's model of SA.
Comprehension
Integrate different sources of
data and organize them to
form a holistic picture of the
environment
Projection
“Project the future actions of
the elements in the
environment”
Decision
Make a decision using an
effective problem-solving
strategy
Intelligent Machines
2019-06-23
6
Perception
Acquire data via sensors
databases, etc.
Comprehension
Use data fusion and machine
learning techniques to
characterize situations and
understand the environment
Projection
Use logical reasoning,
planning, abstract thinking
and other cognitive
techniques to project future
states and dynamics.
Decision
Make a decision using an
effective problem-solving
strategy
Techniques
2019-06-23
7
Big Data
Analytics
Computer
Vision
Natural
Language
Processing
Speech
Recognition
Planning and
OptimizationRobotics
How can AI change our future?
AI
Transportation
Healthcare
Manufacturing
Finance
Agriculture
Media
Retail
Consumer
Electronics
2019-06-23
8
Example applications
• Using predictive analytics to gain insights of customers, sales, and product
performance and provide recommendations on actions that improve companies’
sales.
• Predicting repairs and maintenance needs of devices, machinery, cars, trucks, etc.
• Home Robots
• Self-driving cars
• Disease detection, diagnosis, treatment and management processes.
• Financial search engines and automated financial investing platforms.
• Suggest people to follow, tweets and news based on user’s individual
preferences.
• ……. And the list goes on
2019-06-23
9
Examples from CIE 407 and CIE 555
CIE 407
• The course provides an overview of underlying concepts and practical algorithms
used in machine learning. Topics include linear regression, decision trees, support
vector machines, neural networks, ensembles, and reinforcement learning.
CIE 555
• The course provides an introduction to deep Learning. Topics include
optimization techniques for deep networks, regularization techniques, CNN,
RNNs, LSTM, , and more
2019-06-23
10
Example 1: RL Agent to play Flappy Bird game
• By Esraa Magdy
• ML Technique: Deep Reinforcement
Learning
2019-06-23
11
Example 2: Pneumonia X-Ray Classification
• By: Ahmed Wael
• ML Technique: CNN
2019-06-23
12
Example 3: Banknote Classification
• By:
Mohamed Megahed
Mohamed Gad
Abdallah Shawky
AbdelMoez Elsadany
• Technique: CNN
2019-06-23
13
The future of transportation system
2019-06-23
14
Forces of change: The future
of mobility
Part of a Deloitte series on
the future of mobility
Self-driving Cars
2019-06-23
15
Photo by Vjeran Pavic / The Verge
• 3-D positioning and motion
estimation of surrounding objects
• Accurate detection and prediction
of the motion trajectories of
surrounding objects
• Detect distracted drivers
• Maneuvering and Control
Self-driving cars
2019-06-23
16
Source: Mercedes - dailymail.co.uk
2019-06-23
17
•Rear cross-traffic alert
•Forward collision warning
•Automatically adjusts the headlights in
response to darkness
•Blind Spot Detection and Warning Systems
Safety Applications
Traffic Management
• Use a combination of cameras
and sensors to analyze road
congestion, provide alternative
routes, and plan better road
networks.
• Smarter traffic lights that extends
green intervals when needed.
• Smarter ramp metering solutions
• Smarter incident detection and
management
2019-06-23
18
Egypttoday.com
Traffic Monitoring
2019-06-23
19
A systematic mechanism using time
series analysis that imputes missing
data in large traffic datasets
collected at different locations and
at different times.
“Automatic Imputation of Missing Highway Traffic Volume Data”,
2018, IEEE International Conference on Pervasive Computing and
Communications
Parking
• Better space management
• Better payment and pricing
solutions (payments can be
integrated with other services
such as public transport passing)
• Efficient demand management
• Adding services (electric-vehicle
charging stations, security
features,etc)
2019-06-23
20
www.businessinsider.com
Telematics
2019-06-23
21
https://www.cmtelematics.com/drivewell/
• Monitor vehicles and offer
predictive maintenance
capabilities.
• Provide real-time
information about expected
arrival times.
• Monitor driver behaviour
and offer recommendations
to reduce fuel consumption
and improve safety.
Other applications
• Freight applications: increased delivery speeds, lower shipping price, easy
returns, increased utilization of carriers, enhanced visibility of shipments.
• Transit applications: Increased fleet utilization, better scheduling mechanisms,
better incident management, autonomous buses
• …
2019-06-23
22
Automate sharing data across different systems
• Representation of complex
relationships among cyber and
physical components in integrated
smart mobility applications using a
machine understandable language
2019-06-23
23
Things to consider when implementing AI solutions.
A major concern that should be addressed in your
solution.
Data Privacy
Integrate security into the software development life
cycle
Cybersecurity
Who owns the data if multiple parties are involved
Data Ownership
World-wide adoption of regulations and standards
facilitates cross-border connectivity and collaboration
Regulations and
standards
2019-06-23
24
Questions
melshneawy@zewailcity.edu.eg
2019-06-23
25

Applications of Artificial Intelligence in Transportation Systems

  • 1.
    Applications of ArtificialIntelligence in Transportation Systems: An Overview Dr. Mohamed Elshenawy Assistant Professor, Zewail City of Science and Technology 2019-06-23
  • 2.
    Overview • Why weneed AI? • What are intelligent machine? • How can AI change our future? Transportation as an example. • Things to consider when implementing AI solutions. 2019-06-23 2
  • 3.
    Why we needintelligent machines? •Empower humans with technology and qmplify our human intelligence • Ability to analyze large volumes of data to provide a basis for various human decisions. • Transform the ways we do things and drive more value. • Increased productivity and ability to automate repetitive tasks. • Higher Efficiency. 2019-06-23 3
  • 4.
    How humans assessdifferent situations and make decisions? 2019-06-23 4 Dr. Mica Endsley's model of SA.
  • 5.
    How humans assessdifferent situations and make decisions? 2019-06-23 5 Perception “Perceive the status, attributes, dynamics of relevant elements in the environment” Endsley's model of SA. Comprehension Integrate different sources of data and organize them to form a holistic picture of the environment Projection “Project the future actions of the elements in the environment” Decision Make a decision using an effective problem-solving strategy
  • 6.
    Intelligent Machines 2019-06-23 6 Perception Acquire datavia sensors databases, etc. Comprehension Use data fusion and machine learning techniques to characterize situations and understand the environment Projection Use logical reasoning, planning, abstract thinking and other cognitive techniques to project future states and dynamics. Decision Make a decision using an effective problem-solving strategy
  • 7.
  • 8.
    How can AIchange our future? AI Transportation Healthcare Manufacturing Finance Agriculture Media Retail Consumer Electronics 2019-06-23 8
  • 9.
    Example applications • Usingpredictive analytics to gain insights of customers, sales, and product performance and provide recommendations on actions that improve companies’ sales. • Predicting repairs and maintenance needs of devices, machinery, cars, trucks, etc. • Home Robots • Self-driving cars • Disease detection, diagnosis, treatment and management processes. • Financial search engines and automated financial investing platforms. • Suggest people to follow, tweets and news based on user’s individual preferences. • ……. And the list goes on 2019-06-23 9
  • 10.
    Examples from CIE407 and CIE 555 CIE 407 • The course provides an overview of underlying concepts and practical algorithms used in machine learning. Topics include linear regression, decision trees, support vector machines, neural networks, ensembles, and reinforcement learning. CIE 555 • The course provides an introduction to deep Learning. Topics include optimization techniques for deep networks, regularization techniques, CNN, RNNs, LSTM, , and more 2019-06-23 10
  • 11.
    Example 1: RLAgent to play Flappy Bird game • By Esraa Magdy • ML Technique: Deep Reinforcement Learning 2019-06-23 11
  • 12.
    Example 2: PneumoniaX-Ray Classification • By: Ahmed Wael • ML Technique: CNN 2019-06-23 12
  • 13.
    Example 3: BanknoteClassification • By: Mohamed Megahed Mohamed Gad Abdallah Shawky AbdelMoez Elsadany • Technique: CNN 2019-06-23 13
  • 14.
    The future oftransportation system 2019-06-23 14 Forces of change: The future of mobility Part of a Deloitte series on the future of mobility
  • 15.
    Self-driving Cars 2019-06-23 15 Photo byVjeran Pavic / The Verge • 3-D positioning and motion estimation of surrounding objects • Accurate detection and prediction of the motion trajectories of surrounding objects • Detect distracted drivers • Maneuvering and Control
  • 16.
  • 17.
    2019-06-23 17 •Rear cross-traffic alert •Forwardcollision warning •Automatically adjusts the headlights in response to darkness •Blind Spot Detection and Warning Systems Safety Applications
  • 18.
    Traffic Management • Usea combination of cameras and sensors to analyze road congestion, provide alternative routes, and plan better road networks. • Smarter traffic lights that extends green intervals when needed. • Smarter ramp metering solutions • Smarter incident detection and management 2019-06-23 18 Egypttoday.com
  • 19.
    Traffic Monitoring 2019-06-23 19 A systematicmechanism using time series analysis that imputes missing data in large traffic datasets collected at different locations and at different times. “Automatic Imputation of Missing Highway Traffic Volume Data”, 2018, IEEE International Conference on Pervasive Computing and Communications
  • 20.
    Parking • Better spacemanagement • Better payment and pricing solutions (payments can be integrated with other services such as public transport passing) • Efficient demand management • Adding services (electric-vehicle charging stations, security features,etc) 2019-06-23 20 www.businessinsider.com
  • 21.
    Telematics 2019-06-23 21 https://www.cmtelematics.com/drivewell/ • Monitor vehiclesand offer predictive maintenance capabilities. • Provide real-time information about expected arrival times. • Monitor driver behaviour and offer recommendations to reduce fuel consumption and improve safety.
  • 22.
    Other applications • Freightapplications: increased delivery speeds, lower shipping price, easy returns, increased utilization of carriers, enhanced visibility of shipments. • Transit applications: Increased fleet utilization, better scheduling mechanisms, better incident management, autonomous buses • … 2019-06-23 22
  • 23.
    Automate sharing dataacross different systems • Representation of complex relationships among cyber and physical components in integrated smart mobility applications using a machine understandable language 2019-06-23 23
  • 24.
    Things to considerwhen implementing AI solutions. A major concern that should be addressed in your solution. Data Privacy Integrate security into the software development life cycle Cybersecurity Who owns the data if multiple parties are involved Data Ownership World-wide adoption of regulations and standards facilitates cross-border connectivity and collaboration Regulations and standards 2019-06-23 24
  • 25.

Editor's Notes

  • #4 London terror attack: Uber slammed for being slow to turn off ‘surge pricing’ after rampage https://money.cnn.com/2017/06/04/technology/uber-london-attack-surge-pricing/index.html
  • #6 Perception: a pilot would perceive elements such as aircraft, mountains, or warning lights along with their relevant characteristics (color, speed, size, and location)