The Science of Algorithmic Map Inference

Abstract:

A necessary condition for autonomous vehicles to become “mainstream” is the availability of highly accurate and updatable geographical road network maps. Several large commercial mapmaking efforts by automobile manufacturers and technology companies have been recently announced . The race to build the most accurate maps is “truly on.” A societal concern is that, in the near future, the most accurate maps may not be a public good but a property of private stakeholders. A concerted effort is required to democratize mapmaking, and collaborative efforts like OpenStreetMap (OSM) demonstrate the power of the community coming together to create and maintain maps as a public good.

In this tutorial we will review the emerging area of algorithmic map inference (AMI), i.e., the design of algorithms to automatically build and update maps using diverse data sources , primarily GPS data and satellite images. A substantial body of research has now emerged around AMI primarily in KDD and related communities. Thus it is an opportune time to organize the AMI literature in a proper context and introduce it to a wider audience in the research and applications community.

Program:

We will be in ICC Capital Suite Room 2 (Level 3) on 19 August 2018 1:00 pm - 5:00 pm.

Presenters:

  • Sofiane Abbar, Qatar Computing Research Institute (QCRI)
  • Mohammad Alizadeh, Massachusetts Institute of Technology (MIT)
  • Favyen Bastani, Massachusetts Institute of Technology (MIT)
  • Sanjay Chawla, Qatar Computing Research Institute (QCRI)
  • Songtao He, Massachusetts Institute of Technology (MIT)


Contributors:

  • Hari Balakrishnan, Massachusetts Institute of Technology (MIT)
  • Sam Madden, Massachusetts Institute of Technology (MIT)

When:

  • KDD 2018 (August 19th - August 23rd), London, UK

Duration:

  • 3.5 hrs