Projekt

Autonomous Ferry Control using IRL

Udbyder

Vejleder

Sted

København og omegn

1. Introduction

FORCE Technology

FORCE Technology works globally across industries, cultures and markets to identify solutions in these areas and address other challenges for the maritime world.
Our maritime competencies are built on talent and pioneering expertise in maritime technology and services, as we constantly push the boundaries for commerce, competence and cooperation.

The project
Currently (until April 2022) FORCE participate in a project called shippinglab ( https://ship-pinglab.dk/ ). Here the goal is to improve the mathematical model (the digital twin) by using machine learning methods on measurements recorded onboard the Scandline Ferry M/F Copenhagen.
Besides improving the mathematical model of the Ferry, FORCE is also interested in performing autonomous port operations, and using machine learning techniques to derive the energy optimal strategy.

 2. Berthing operations

A Ferry like M/F Copenhagen that operates the Gedser Rostock route needs to rotate 180 degree two times on one of the transits to let the cars drive forward out. On the Gedser – Rostock route the rotations are conducted when the Ferry leaves Gedser, and again just before it goes to berth in Rostock.
Hence the ferry goes astern out of the port in Gedser. Just outside the port the ferry turns around and when in line with the approach channel heads towards Rostock.
The operation is sketched below (see pdf) where the red dotted line illustrates the port and the channel limits.
The blue line is the track recorded from the ferry, and the orange line is obtained by commanding the recorded control settings to the FORCE mathematical model. Yes, the model needs to be improved, and we are working on that.

3. Task
The project consists of deriving a policy/strategy that can be used for autonomous departing the ferry. The derivation should use Inverse Reinforcement Learning on these expert conducted departures, of which FORCE Technology has gigabytes of data for different transits available for the project.
If time allows, the same method could be used on arrival scenarios.
If time allows, the use of Reinforcement Learning to derive the energy optimal usage of the ferrys control units could be attempted.

4. FT deliveries

FT can deliver sensor recordings for a number of departures/arrivals
FT can deliver a simulator and a python interface to control the simulator
FT can participate in status meetings.

I samarbejde med

Force Technology (https://forcetechnology.com/da)

Forudsætninger

An ability and desire to work exploratively in a cross-disciplinary environment. Mathematical maturity.

Download materiale

Søg i opslag
Tags
Kontakt

Virksomhed/organisation

DTU Compute

Navn

Anders Stockmarr

Stilling

Lektor

Mail

anst@dtu.dk

Vejleder-info

Kandidatuddannelsen i Matematisk Modellering og Computing

Vejleder

Anders Stockmarr

ECTS-point

30 - 35

Type

Kandidatspeciale

Skal have taget

02450,02402/02403 or similar. 02417 desirable.

OM DTU

DTU er et teknisk eliteuniversitet med international rækkevidde og standard. Vores mission er at udvikle og nyttiggøre naturvidenskab og teknisk videnskab til gavn for samfundet. 10.000 studerende uddanner sig her til fremtiden, og 5.700 medarbejdere har hver dag fokus på uddannelse, forskning, myndighedsrådgivning og innovation, som bidrager til øget vækst og velfærd.

Find os her

Anker Engelunds Vej 1
Bygning 101A
2800 Kgs. Lyngby


45 25 25 25

Kontakt og besøg

CVR-nr. 30 06 09 46

Liste over EAN Numre

Job på DTU

Se alle jobs
 

loading..