Machine learning-based geocoding & intelligent geofencing

Imagine a situation - You have 1,000 parcels to deliver to unique addresses and your truck has a capacity of 100 parcels. The number of possible combinations from one address to other could be {C(1000,100)}

6.39 × 10^114 (Close to 115 digits!)

A computer would take some days and a human - some years with a lot of coffee to check for the most optimized sequence based on distance and travel time for these many numbers of possibilities.

Routing requires accurate geocoding for both route planning and ETA calculations. FarEye uses sophisticated algorithms to self-learn the ‘correct’ address from attempted delivery addresses, success and failures. FarEye transforms a description of a location—such as a pair of coordinates, an address, or a name of a place—to a location on the earth's surface. The address is spatially displayed on the locations and patterns are recognized within the information. It converts English addresses into geographic coordinates as latitude & longitudes, which are used to mark positions on a map.
 

As next steps, an intelligent geofencing is created in order to improve efficiency.

• The manager can create different geofences of any shape or size, as a guide for auto-routing

• This is typically used to mark out areas that should not be clubbed when creating routes

• Routes are automatically recommended as per the load and segregation in the fences

• The manager can combine multiple geo fences dynamically to optimize for that day’s volume and load

• The manager can monitor unlimited locations, override and tweak the individual points on the route and reassign to different parcels

• FarEye supports manual sticky routing where the manager can assign riders familiar with specific parts of the city/area or zone

All this and a million more algorithms and processes are run simultaneously in the back-end while fetching the most optimized routes for the complete fleet.