Address Matching Using Truck Tours Feedback

TitleAddress Matching Using Truck Tours Feedback
Publication TypeConference Paper
Year of Publication2015
AuthorsBouallouche, D, Vioix, J-B, Millot, S, Busvelle, E
EditorBlecker, T, Kersten, W, Ringle, CM
Title of ProceedingsOperational Excellence in Logistics and Supply Chains
Volume22
Page391
Publisherepubli
Conference LocationHamburg
ISBN Number978-3-7375-4056-8
Other NumbersISSN (print) 2365-4430, ISSN (online) 2365-5070
KeywordsAddress Matching, Driver Experience Feedback, Geocoding Correction, Real Truck Tours
Abstract

When researchers or logistics software developers deal with vehicle routing optimization,
they mainly focus on minimizing the total traveled distance or time of the
tours, and maximizing the number of visited customers. However, in real transporter
situations, the actual data received is often of bad quality, particularly the irrelevance
of addresses and address geocoding errors. Therefore, trying to optimize tours
with impertinent customers' GPS-coordinates, which are the most important input
data for solving a vehicle routing problem, will lead to an incoherent solution, especially
if the locations of the customers used for the optimization are very different
from their real positions.
Our work is supported by a logistics software editor Tedies (2013) and a transport
company Upsilon (2009). We work with the company's real truck routes data to carry
our experiments. The aim of this work is to use the experience of the driver and the
feedback of the real truck tours to validate and correct GPS-coordinates to the next
tours. Our method significantly improves the quality of the geocoding.
This study shows the importance of taking into account the feedback of the trucks to
gradually correct address geocoding errors. Indeed, the accuracy of customer’s address
and its GPS-coordinates plays a major role in tours optimization. This feedback
is naturally and usually taken into account by transporters (by asking drivers, calling
customers, …), to learn about their tours and bring corrections to the upcoming
tours. Hence, we develop a method to do most of that automatically.

DOI10.15480/882.1265