Thesis Student

Use ML learning approaches to generate real-time heterogenous routing and transport mission management of container transports


Background


Transport automation of goods is essential to reduce cost and energy usage. Here in this thesis, we will study container transports for pre-selected fleets, that are optimized for a predicted mission need for driving containers from the sea-port Gothenburg to a dry-port in Viared-Borås and then distributed out the customers Ci


The goal of the thesis is to use machine learning approaches to solve the weekly and realtime planning and execution of the fleet usage for transport. This means setting up a mathematical model for the problem and selecting appropriate solution methods. It also includes to make the Table 1 transport demand stochastic and to include realistic traffic flow for surrounding traffic modelling. The realistic traffic flow has been recorded for one of the routes for about 200 days. The recorded data is available for the thesis. The recorded data shall be used for estimating the traffic density, waiting time at nodes and the type of loading-unloading, at different times of the day.


The thesis work can be summarized as:


1.- Set up a virtual environment for running stochastic goods flow from seaport, dryport and customers, with traffic simulations and vary the traffic density, e.g. using Sumo as training environment. An example of using Sumo can be found in [2], and the paper include github reference.
2.- Investigate the feasibility of using Machine Learning (ML) [3] approaches mixed with classical optimization for solving the realtime optimization of the stochastic transport mission.
3.- Use a classical optimization formulation to only validate the solutions from ML are feasible and cost-efficient.


Purpose
The aim of this thesis is to use advanced machine learning methods to solve problems within autonomous driving. The work will include programming, machine learning, control theory, numerical optimization and vehicle simulation.


Additional Information
The work will be carried out at Volvo Group Trucks Technology, Sweden. The thesis is recommended for one or two students with a strong background in machine learning and Python/C++ with a good mathematical background. Prior experience with control theory and modeling/simulation is meritorious.
Location: Preferably Göteborg, Sweden
Time Schedule: Preferably Jan 2023- Jun 2023, but we are flexible to adapt to your time schedule


Contact Persons


Ingmar Bengtsson – Volvo GTT
email: ingmar.bengtsson@volvo.com


Leo Laine – Volvo GTT
email: leo.laine@volvo.com


Gabriel Ibanez – Volvo GTT
email: gabriel.ibanez@volvo.com


Bibliography


See the list of bibliography for related articles.


[1] T. Ghandriz, B. J. H. Jacobson, M. Islam, J. Hellgren, and L. Laine, ‘Transportation-mission-based Optimization of Heterogeneous Heavy-vehicle Fleet Including Electrified Propulsion’, Energies, vol. 14, no. 11, 2021, doi:10.3390/en14113221. URL: https://www.mdpi.com/1996-1073/14/11/3221


[2] Carl-Johan E Hoel. “Decision-Making in Autonomous Driving using Reinforcement Learning”. PhD thesis. Chalmers Tekniska Högskola (Sweden),2021. URL:https://research.chalmers.se/publication/526543/file/526543_Fulltext.pdf


[3] Chi, C., Aboussalah, A. M., Khalil, E. B., Wang, J., & Sherkat-Masoumi, Z. (2022). A Deep Reinforcement Learning Framework For Column Generation. arXiv preprint arXiv:2206.02568. URL: https://arxiv.org/abs/2206.02568