Deadlock prevention of autonomous vehicle fleets

A fleet of autonomous vehicles needs commanding for its execution. Tactical decisions, with a typical horizon in the order of minutes, are handled by a traffic planner. A vital input to the planner is the current vehicle positions. In broad sense, the planner shall maximize fleet productivity with respect of a set of constraints.
One obvious constraint is that two vehicles shall not collide. Another, less obvious, constraint is that deadlock situations shall be avoided.


In the figure below is a deadlock present. Only one vehicle can be in the area named SL, i.e. only one of the nodes 4, 5, 8 and 9 can be occupied. The consequence is that none of the three vehicles can move because all waits for another vehicle to move.







This master thesis handles the following questions:


How can a deadlock situation be expressed mathematically?
How can a traffic planner agent be informed that a state, set of vehicle positions, is a deadlock?
How can the questions above be answered in a computationally efficient manner?



The thesis work will include various fields such as machine learning, simulation and optimization. Personal interest in programming is seen as benefit. The work will be carried out at Volvo Autonomous Solutions in Gothenburg. The thesis is recommended for one student with programming profile and good mathematical skills.


Don’t hesitate to contact our manager, Dominika Wroblewska, Manager Cloud, +73-902 72 94

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


Thesis Level: Master

Language: English


Starting date: 9/01/2023


Number of students: 1-2


Tutor
Jonas Hellgren, Principal Engineer, jonas.hellgren@volvo.com