Dela med andra

Collision Avoidance for Marine Vessels using Deep Reinforcement Learning

Publicerad 2021-10-25


Master thesis
----------------
Description
....................
Developing autonomous collision avoidance systems for marine vessels that can operate in an unstructured and unpredictable environment is challenging. Particularly in congested sea areas, each ship should make decisions continuously to avoid collisions with other ships in a busy and complex waterway. The main task of this project is to propose an efficient method based on deep reinforcement learning to solve collision avoidance problem for marine vessels. 


Arbetsuppgifter
Scope
-----------
Research topics:
How to guarantee safety for marine vessels in the presence of static and moving obstacles?
How to account for marine traffic rules, speed limitations and vessel's dynamics?
Goal(s): 
Explore how reinforcement learning can be applied to collision avoidance for ships
Generate paths which are collision free and satisfy kinematics and dynamics constraints 

Approach
------------
The work will address the following points: 
Prior art review on path planning and collision avoidance for marine vessels 
Formulate the collision avoidance design as a reinforcement learning (RL) problem
Design an algorithm to solve the RL problem
Validate for several simulated scenarios


Kvalifikationer
Knowledge on Control Theory
Basic knowledge on Machine Learning
Good implementation skills
Knowledge on  reinforcement learning is a plus 


Kontaktperson
Hamid Feyzmahdavian
hamid.feyzmahdavian@se.abb.com
+46705427436

Jobbansökan
hamid.feyzmahdavian@se.abb.com


Tillbaka
  Arbetsgivare
ABB AB

Antal platser
1-2 st


Ort
Västerås

Sista ansökningdag
2022-02-23

Tidsperiod
Master thesis



Bookmark and Share