Foto del docente

Roberto Baldacci

Associate Professor

Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi"

Academic discipline: MAT/09 Operations Research

Teaching

Dissertation topics suggested by the teacher.

Theses available with BLUEROCK logistics.

BLUEROCK logistics: Company Profile

We support logistics service providers and shippers to simplify their supply chain. We are on
the fore-front of technology development and active in the scientific community of data science
and logistics. Our proposition is state-of-the-art. With an App based architecture, modern
screen design and our expertise in logistics, software as well as in modeling/optimizing we
draw the attention of many industries.
Our main customers are in France, Germany, Italy, Benelux and other European countries. In
addition, we are expanding in North-America. Eventually we need to support our customers
globally.
Requisites (in addition to what is stated in the projects description):
- Fluency in English
- Good knowledge and interest in optimization issues and algorithms applied to logistics
- Strong analytical skills
Workplace: ‘s-Hertogenbosch The Netherlands
Economic conditions and benefits: reimbursement of the cost of the accommodation in the
Netherland plus an allowance (budget to be agreed).
If the project ends with mutual satisfaction, the next step might be the offer of an employment
contract.
Company website: www.bluerocklogistics.com

In particular, the following topics are of interest.

Highly flexible FTL/LTL vehicle routing

The aim is to create a generic framework for vehicle routing problems with only few stops
per trip. The framework should also include visualization on a map for standalone use.
The main idea behind the algorithm is very simple. We first enumerate all possible trips,
after which we run a set partitioning solver to get the best combination of trips. The main
challenge in this lies within the enumeration of the possible trips. We need to consider
many different constraints such as time windows, capacity, single delivery, etc. These
practical issues make it quite difficult to develop a good generic framework.
Moreover, the cost function for the set partitioning to work needs to be very well balanced.
However, the actual routing cost are often depending on weight and/or distance which can
force (accurate) estimations to be used instead of the actual cost.
For further extension we would also like to investigate the possibility of including the time
aspect into the problem. At this moment it is unclear whether this requires an additional
step or we can fit it in the same framework.
Programming language: Java and/or Julia (preferred)
Deliverables: generic framework, tools for evaluation/analysis

Fast estimation and calculation of distance matrices

The aim is to create a generic framework for distance/travel time calculations. We use
distances and travel times in many different situations but we do not always need the same
accuracy. In case we are planning trips over whole of Europe, we might be ok with distances
between postal codes, whereas when we are planning a city trip we need very detailed
information.
The main idea is to create a generic service which will validate and geocode the addresses
for its specific purpose, followed by generation of the matrix. In addition, the end-user has
certain constraints on the vehicle type, border crossing, toll, etc., because sometimes the
fastest route is not the preferred route.
Moreover, validation of the distances and travel times is also very important. Since many
matrices will be used directly in algorithms it because incredibly hard for the end-user to
find mistakes afterwards.
For further extension, we would like to investigate comparison methods between different
approaches to improve distance estimations which would speed up the calculations even
further.
Programming language: Java (preferred)
Deliverables: generic framework, tools for validation

Automatic generation of micro zones

A micro zone is a coherent small geographical area that has on average a particular
distribution work load over a time period. For example, on average 30 minutes working time
(inter-stop driving and stop times). Finding volumes to combine will not be that difficult but
how to detect natural barriers like channels, highways, etc.
The aim is to create a framework in which micro zones can be generated and adjusted. It
will be completely standalone since only the output will be further used. The main challenge
would be combining the geographical data with the algorithm output to form the basis for
better and improved micro zones.
This is experimental research where we combine algorithms, GIS-technology and statistics.
Programming language: Any suitable language
Deliverables: standalone framework

Forecasting with user feedback

We get more data every day and we would like to predict the future. However, sometimes
there are events that happen which we cannot predict, but can be specified by a human.
This is exactly the area in which we can improve. The aim is to create a standalone
framework which helps the end-user to make a supervised forecast of the future.
The main challenge is in visualizing and processing the data in different ways to make good
forecasts which are useful for further use. Especially think about aggregating data in
different slices such that we get accurate forecasts which then need to be translated back
into meaningful data points.
Moreover, end-users should have a big influence on special events that have an impact on
the forecast. If these are inputted by the user upfront, we can take them into consideration.
Programming language: R or Python
Deliverables: standalone framework

 

Latest news

At the moment no news are available.