IJCAI 2013 Tutorial on AI-Driven Analytics In Traffic Management

Analytics is the backbone for intelligence core in traffic management. It can be varied ranging from how to move people and goods through best routes on available modes (vehicles) under different settings, coordination of lights or setting up a cost-effective sensing infrastructure. The aim of the tutorial is to present AI-enabled analytics related to transportation in a consolidated manner from varied domains like transportation, social networks, planning and graph theory to early and experienced researchers. With this, we hope that results from different areas can be better reused leading to increased attention of the AI-community enabling smarter traffic management.

Where: 23rd International Joint Conference on Artificial Intelligence, August 3-9, Beijing, China (http://ijcai13.org/)
When:  Saturday, August 3rd, 2013; Morning half (http://ijcai13.org/program/tutorial/TA2)
DownloadPresentation
          
History

This tutorial is a subsequent one to a previous tutorial "Traffic Management and AI", in conjunction with 26th Conference on Artificial Intelligence (AAAI-12), at Toronto, Canada, July 22-26, 2012 conducted by one of the current organizers.  More details can be found here: http://www.aaai.org/Conferences/AAAI/2012/aaai12tutorials.php.

Tutorial Description

Traffic management is a pressing problem for cities around the world. Moreover, it is a highly visible perspective of a city's life affecting all aspects of its citizens' economic and personal activities. Consequently, there is substantial academic and commercial interest in addressing this problem. 

AI can contribute significantly to analytics in traffic management.  A common concern in traffic management is to find ways to move people and goods through best routes. However, the nature of the actual problem changes based on the setting---route has to be found for one agent v/s some v/s all, how the travel times change, whether the vehicles move at scheduled time v/s opportunistically, environmental concerns. Further, techniques related to path planning have been looked in other domains like robotics, social networks, planning and graph theory. Other analytics relate to traffic light coordination and traffic sensing. If this information could be presented in a consolidated manner, it would lead to better reuse of known results, thus leading to increased attention of the AI community to pressing problems in traffic management.

The topics we intend to cover in the tutorial include the following:
  • Traffic management ecosystem overview
    • Basics concepts: demand, supply, congestion    
    • Key performance indicators
    • Core traffic issues 
    • Sensing technologies and their implications for knowledge representation 
  • Problem settings for traffic analytics 
    • Travel planning
    • Traffic light coordination 
    • Sensing optimization 
  • Travel planning
    • Setting---Models of transportation world,  Type of traveler: single traveler (agent) v/s multi traveler (agent), full or partial observation
    • Methods
    • Optimization criterion---shortest time,  optimal fare,  energy or battery consumption,  multimodal route finding 
    • Distribution coordination techniques
  • Traffic light coordination---background, latest trend
  • Sensing optimization
  • Competitions, datasets relevant for traffic analytics
  • Practical considerations
The AI topics we will touch are: 
  • Graph and Path Analytics 
  • Planning 
  • Scheduling 
  • Knowledge representation, ontology 
  • Learning 
  • Data Mining
Background of Presenters
          
          Biplav Srivastava
Senior Researcher & Master Inventor; ACM Senior Member IBM Research India
4, Block - C, Institutional Area
Vasant Kunj, New Delhi - 110 070, India
sbiplav@in.ibm.com

Dr. Biplav Srivastava, Senior Researcher & Master Inventor, IBM Research and ACM Senior Member, is based out of New Delhi, India. Biplav has interests in planning, scheduling, policies, learning and information representation/ontology. He has 90+ research papers published including all top fora in his fields, 20 US Patents issued and 45 applications. Biplav actively participates in professional services globally including running the ‘AI in India’ virtual group, organizing conference tracks, workshops and as a Program Committee member for more than 50 events. More details are at: http://www.research.ibm.com/people/b/biplav/



Akshat Kumar 
Research Scientist
IBM Research India
4, Block - C, Institutional Area
Vasant Kunj, New Delhi - 110 070, India 
akkumaro@in.ibm.com   

Akshat Kumar is a research scientist at IBM research India working with the analytics and optimization group. He has received PhD in computer science from University of Massachusetts Amherst in 2012. His main research interests are in planning under uncertainty, multiagent systems, machine learning and optimization. He has a keen interest in solving real-world inspired problems using planning and optimization based techniques. His current focus is on the emerging field of computational sustainability. He has published in major AI and ML conferences such as AAAI, IJCAI, NIPS and UAI. He is also the recipient of UMass graduate school fellowship for the year 2010-11.   More details are at: http://people.cs.umass.edu/~akshat/