LastMile:

Rethinking last-mile delivery through crowdsourcing

 

Open call for post-doctoral grant: https://www.inesctec.pt/ip-en/work-with-us/bolsas-inescporto/concurso-para-a-atribuicao-de-1-bolsa-de-pos-doutoramento-projeto-XXXXXXXXX   


Last Mile Delivery

Delivery of products purchased by online customers is an increasingly important problem for e-commerce companies, representing a substantial para of their costs.  The final step --- last mile delivery, which conveys the product to the customer --- is a key stage in the process,

Several business models proposed recently exploit the concept of crowdsourcing in which, in addition to professional couriers, ordinary citizens are invited to participate in the delivery process. However, current models present a set of weaknesses and do not adequately address many of the concerns associated with real-world implementation.

This project will address such concerns.  The main concept is to have in-store customers delivering parcels on their way home to customers that ordered on-line, besides maintaining a backup of professionals for delivering the remaining parcels. This approach will reduce environmental impacts and urban congestion at low marginal cost to the company.


Optimization

The underlying problem associated to LastMile can be modeled as a mixed-integer optimization problem.  The most elementary models are deterministic, but for realistic outcomes they must be extended to stochastic contexts.  We expect that the solution process will involve a intricate framework, making use of optimization solvers as well as ad hoc components specific to this problem.


Machine learning

An important part of the problem involves modeling the probability of acceptance for outsourcing tasks, as well as their impacts on delivery costs.  This will involve dynamically analyzing data and extracting relevant information, making use of the state-of-the-art in machine learning.


Steps

  1. Modeling practically relevant situations yet unaddressed

  2. Researching new compensation schemes for guaranteeing sustainability of the models

  3. Researching efficient optimization algorithms for simultaneously matching parcels to couriers and optimizing the routes of the firm’s own fleet.


Keywords:

  1. Vehicle routing

  2. Transportation

  3. Stochastic programming

  4. Machine learning


Key competences:

  1. Mathematical modeling and optimization

  2. Stochastic modeling

  3. Computer programming



Project team

João Pedro Pedroso - http://orcid.org/0000-0003-1298-7191

Ana Viana - http://orcid.org/0000-0001-5932-5203

Xenia Klimentova - http://orcid.org/0000-0003-1085-0810

Pierre Hosteins - http://orcid.org/0000-0003-4186-9127

Katarzyna Gdowska - http://orcid.org/0000-0002-7964-3724

Abdur Rais - http://orcid.org/0000-0003-1906-9684