Msc Thesis: ML for combinational behavior modeling of a truck

Driving automation has a significant potential in improving the traffic safety by reducing the human errors. Volvo Group highly focuses on the development of Advanced Driver Assistance System and Autonomous Trucks. Identifying various traffic driving scenarios that the trucks are exposed to and training the autonomous system to take safest and efficient actions in those scenarios are crucial. For this purpose, we need to develop realistic traffic environments that consists of realistic surrounding vehicle models.
In this project, we aim to explore a real-world traffic dataset available at Volvo, to learn the combinational behavior patterns of a truck and surrounding vehicles. For instance, a passenger car driver may not feel safe and try to change lane or slow down when the lateral gap with a truck in the next lane is too small. Our goal is to develop machine learning models to capture such subtle behaviors.


The purpose of this thesis can be summarized as:
Perform data analysis and preprocessing on the real-world traffic dataset surrounding a truck and extract various traffic scenarios.
Learn the combinational behavior patterns of the vehicles using supervised learning algorithms and/or reinforcement learning. For example, how the surrounding passenger cars respond when the ego vehicle turns on the left/right indicators or when the ego vehicle is closer to the lane marker, where is the longitudinal /lateral position and speed of passenger car relative to the truck combination etc.
Model the surrounding vehicles using the learned behavior patterns in a simulation platform such as SUMO.

This project will use advanced machine learning methods to solve problems within autonomous driving. The work will include data preprocessing, machine learning, and vehicle simulation. 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 data science, machine learning and Python/C++ with a good mathematical background.


Thesis Level: Master

Language: English


Starting date: 01-01-2024


Number of students: 1-2


Tutor
Leo Laine – Volvo GTT
Deepthi Pathare – Volvo GTT


Kindly note that due to GDPR, we will not accept applications via mail. Please use our career site.