Overview of Cognitive Routing

The Cognitive Routing function learns from historical data and predicts the best route for you.

Use case examples

Fleet management and others benefit from the following Cognitive Routing use cases.

• Register the delivery locations as POIs.
• Estimate arrival time for each delivery location with current traffic data.
• Predict the route of subcontracted driver from historical probe data.

Key features

• The points of interest (POI) data support, which each customer can customize via the POI API.
• Integration with external data, which is provided from a third party via a Data Crawler. Examples of third parties include: The Weather Company, TomTom traffic, The Weather Company weather.
• MPP&DP (Most Probable Path & Destination Prediction) with the machine learned model data from the historical probe data.
• Enhanced route service in combination with the following data.
  • More road link attribute data from a map vendor such as average speed, road class and other data.
  • The pattern data learned from the historical probe data per vehicle or per driver.
  • The real-time traffic flow data from a third party (The Weather Company and TomTom).
  • The POI data.

Architecture diagram

Summary of functions

• POI API
  • create/update
  • delete
  • get
  • query
    • value filter support
    • circle (radius) support
• Route API
  • Route search and arrival time estimation
    • Multiple way points (Longitude/Latitude list or POI list)
    • Shortest distance route
    • Shortest time route
    • Optimization of route with average speed (only if TomTom map data is used)
  • Route prediction
    • Route prediction per vehicle or per driver (* Pre-requisite: MPP&DP's trajectory pattern data)
• MPP&DP API
  • Generate prediction model per vehicle (per moving object) or per driver
  • Predict destination and route with current trip data
  • Predict destination and route with passed probe data