Learning and routing in last-mile delivery

The vehicle routing problem and its extensions have been successful in optimizing a measure such as time or cost, but only when these measures can be accurately estimated or approximated. However, this is not the case in complex environments. Particularly in urban routing, it is generally complicated, and efficient routing requires tacit knowledge of the drivers. That is the main reason for which the drivers may change the routes that are ‘optimized’ by routing algorithms during the operations. The routes generated by automated systems are also generally modified by experienced staff for optimization purposes. The objective of the project is to capture such tacit knowledge in the complex urban environments, and to incorporate this information along with the changes in operational procedures into routing algorithms, to minimize the human interventions a posteriori to generation of the routes.

Faculty Supervisor:

Okan Arslan

Student:

Partner:

Intelcom

Discipline:

Engineering

Sector:

Transportation and warehousing

University:

HEC Montréal

Program: