# Marvin Kastner, M.Sc.

#### Adresse

Technische Universität Hamburg

Institut für Maritime Logistik

Am Schwarzenberg-Campus 4 (D)

21073 Hamburg

#### Kontaktdaten & Profile

Büro: Gebäude D Raum 5.007

Tel.: +49 40 42878 4793

E-Mail: marvin.kastner(at)tuhh(dot)de

ORCiD: 0000-0001-8289-2943

LinkedIn: https://www.linkedin.com/in/marvin-kastner/

ResearchGate: https://www.researchgate.net/profile/Marvin-Kastner

Google scholar: https://scholar.google.de/citations?user=lAR-oVAAAAAJ&hl=de&oi=ao

Scopus: https://www.scopus.com/authid/detail.uri?authorId=57221938031

#### Forschungsschwerpunkte

- simulationsgestütztes Planen von Container-Terminals
- Optimierung der Ablaufplanung im Yard von Container-Terminals
- technologiegestützte Verbesserung der maritimen Sicherheit
- Maschinelles Lernen in der maritimen Logistik
- Optimierung multivariater Black-box Funktionen

#### Vorträge und Workshops (Auszug)

- 06.05.2024 ein Workshop an der Graduiertenakademie der TUHH: "Einführung in Jupyter Notebooks" [mehr]
- 25.01.2023 ein Vortrag auf dem 7. Suderburger Logistik-Forum: "KI-unterstützte Planung von Güterumschlaganlagen am Beispiel von Containerterminals"
- 15.09.2022 ein Vortrag bei den MLE-Days 2022: "Synthetische Daten für das Reinforcement-Learning bei Container-Terminal-Steuerungen"
- 28.06.2022 ein Workshop an der Graduiertenakademie der TUHH: "Einführung in Jupyter Notebooks" [mehr]
- 02.07.2021 ein Workshop bei den MLE-Days 2021: "Methoden des Maschinellen Lernens in der Maritimen Logistik" [zip]
- 16.03.2021 ein Workshop an der Graduiertenakademie der TUHH: "Einführung in Jupyter Notebooks" [mehr]
- 30.11.2020 im Rahmen der Vortragsreihe "Train Your Engineering Network" der MLE-Initiative: "How to Talk About Machine Learning with Jupyter Notebooks" [mehr]
- 22.11.2019 auf der DISRUPT NOW! AI for Hamburg: "Künstliche Intelligenz in der maritimen Wirtschaft" [mehr]
- 29.10.2019 im Rahmen der forschungsbörse: "Maritime Logistik - Ein Rundumschlag" [mehr]
- 23.10.2019 bei der Open Access Week 2019 an der TUHH: "Datenanalyse - Offener Workshop: Daten auswerten und visualisieren mit Jupyter Notebooks" [mehr] [git]
- 16.11.2018 beim GI DevCamp Hamburg: "Mobility Research and GDPR"
- 27.09.2018 beim SGKV AK zum Thema Lkw-Ankünfte: "Prognoseverfahren und neuronale Netze – Was ist möglich?"

#### Veröffentlichungen (Auszug)

#### 2024

[182412] |

Title: Robust berth scheduling using machine learning for vessel arrival time prediction. |

Written by: Kolley, Lorenz and Rückert, Nicolas and Kastner, Marvin and Jahn, Carlos and Fischer, Kathrin |

in: <em>Flexible Services and Manufacturing Journal</em>. 9 (2022). |

Volume: <strong>35</strong>. Number: |

on pages: 29-69 |

Chapter: |

Editor: |

Publisher: |

Series: |

Address: |

Edition: |

ISBN: |

how published: |

Organization: |

School: |

Institution: |

Type: |

DOI: 10.1007/s10696-022-09462-x |

URL: https://doi.org/10.1007/s10696-022-09462-x |

ARXIVID: |

PMID: |

**Note: **i3lab

**Abstract: **In this work, the potentials of data-driven optimization for the well-known berth allocation problem are studied. The aim of robust berth scheduling is to derive conflict-free vessel assignments at the quay of a terminal, taking into account uncertainty regarding the actual vessel arrival times which may result from external influences as, e.g., cross wind and sea current. In order to achieve robustness, four different Machine Learning methods-from linear regression to an artificial neural network-are employed for vessel arrival time prediction in this work. The different Machine Learning methods are analysed and evaluated with respect to their forecast quality. The calculation and use of so-called dynamic time buffers (DTBs), which are derived from the different AIS-based forecasts and whose length depends on the estimated forecast reliability, in the berth scheduling model enhance the robustness of the resulting schedules considerably, as is shown in an extensive numerical study. Furthermore, the results show that also rather simple Machine Learning approaches are able to reach high forecast accuracy. The optimization model does not only lead to more robust solutions, but also to less actual waiting times for the vessels and hence to an enhanced service quality, as can be shown by studying the resulting schedules for real vessel data. Moreover, it turns out that the accuracy of the resulting berthing schedules, measured as the deviation of planned and actually realisable schedules, exceeds the accuracy of all forecasts which underlines the usefulness of the DTB approach

##### 2023

[182412] |

Title: Robust berth scheduling using machine learning for vessel arrival time prediction. |

Written by: Kolley, Lorenz and Rückert, Nicolas and Kastner, Marvin and Jahn, Carlos and Fischer, Kathrin |

in: <em>Flexible Services and Manufacturing Journal</em>. 9 (2022). |

Volume: <strong>35</strong>. Number: |

on pages: 29-69 |

Chapter: |

Editor: |

Publisher: |

Series: |

Address: |

Edition: |

ISBN: |

how published: |

Organization: |

School: |

Institution: |

Type: |

DOI: 10.1007/s10696-022-09462-x |

URL: https://doi.org/10.1007/s10696-022-09462-x |

ARXIVID: |

PMID: |

**Note: **i3lab

**Abstract: **In this work, the potentials of data-driven optimization for the well-known berth allocation problem are studied. The aim of robust berth scheduling is to derive conflict-free vessel assignments at the quay of a terminal, taking into account uncertainty regarding the actual vessel arrival times which may result from external influences as, e.g., cross wind and sea current. In order to achieve robustness, four different Machine Learning methods-from linear regression to an artificial neural network-are employed for vessel arrival time prediction in this work. The different Machine Learning methods are analysed and evaluated with respect to their forecast quality. The calculation and use of so-called dynamic time buffers (DTBs), which are derived from the different AIS-based forecasts and whose length depends on the estimated forecast reliability, in the berth scheduling model enhance the robustness of the resulting schedules considerably, as is shown in an extensive numerical study. Furthermore, the results show that also rather simple Machine Learning approaches are able to reach high forecast accuracy. The optimization model does not only lead to more robust solutions, but also to less actual waiting times for the vessels and hence to an enhanced service quality, as can be shown by studying the resulting schedules for real vessel data. Moreover, it turns out that the accuracy of the resulting berthing schedules, measured as the deviation of planned and actually realisable schedules, exceeds the accuracy of all forecasts which underlines the usefulness of the DTB approach

##### 2022

[182412] |

Title: Robust berth scheduling using machine learning for vessel arrival time prediction. |

Written by: Kolley, Lorenz and Rückert, Nicolas and Kastner, Marvin and Jahn, Carlos and Fischer, Kathrin |

in: <em>Flexible Services and Manufacturing Journal</em>. 9 (2022). |

Volume: <strong>35</strong>. Number: |

on pages: 29-69 |

Chapter: |

Editor: |

Publisher: |

Series: |

Address: |

Edition: |

ISBN: |

how published: |

Organization: |

School: |

Institution: |

Type: |

DOI: 10.1007/s10696-022-09462-x |

URL: https://doi.org/10.1007/s10696-022-09462-x |

ARXIVID: |

PMID: |

**Note: **i3lab

**Abstract: **In this work, the potentials of data-driven optimization for the well-known berth allocation problem are studied. The aim of robust berth scheduling is to derive conflict-free vessel assignments at the quay of a terminal, taking into account uncertainty regarding the actual vessel arrival times which may result from external influences as, e.g., cross wind and sea current. In order to achieve robustness, four different Machine Learning methods-from linear regression to an artificial neural network-are employed for vessel arrival time prediction in this work. The different Machine Learning methods are analysed and evaluated with respect to their forecast quality. The calculation and use of so-called dynamic time buffers (DTBs), which are derived from the different AIS-based forecasts and whose length depends on the estimated forecast reliability, in the berth scheduling model enhance the robustness of the resulting schedules considerably, as is shown in an extensive numerical study. Furthermore, the results show that also rather simple Machine Learning approaches are able to reach high forecast accuracy. The optimization model does not only lead to more robust solutions, but also to less actual waiting times for the vessels and hence to an enhanced service quality, as can be shown by studying the resulting schedules for real vessel data. Moreover, it turns out that the accuracy of the resulting berthing schedules, measured as the deviation of planned and actually realisable schedules, exceeds the accuracy of all forecasts which underlines the usefulness of the DTB approach

##### 2021

[182412] |

Title: Robust berth scheduling using machine learning for vessel arrival time prediction. |

in: <em>Flexible Services and Manufacturing Journal</em>. 9 (2022). |

Volume: <strong>35</strong>. Number: |

on pages: 29-69 |

Chapter: |

Editor: |

Publisher: |

Series: |

Address: |

Edition: |

ISBN: |

how published: |

Organization: |

School: |

Institution: |

Type: |

DOI: 10.1007/s10696-022-09462-x |

URL: https://doi.org/10.1007/s10696-022-09462-x |

ARXIVID: |

PMID: |

**Note: **i3lab

**Abstract: **In this work, the potentials of data-driven optimization for the well-known berth allocation problem are studied. The aim of robust berth scheduling is to derive conflict-free vessel assignments at the quay of a terminal, taking into account uncertainty regarding the actual vessel arrival times which may result from external influences as, e.g., cross wind and sea current. In order to achieve robustness, four different Machine Learning methods-from linear regression to an artificial neural network-are employed for vessel arrival time prediction in this work. The different Machine Learning methods are analysed and evaluated with respect to their forecast quality. The calculation and use of so-called dynamic time buffers (DTBs), which are derived from the different AIS-based forecasts and whose length depends on the estimated forecast reliability, in the berth scheduling model enhance the robustness of the resulting schedules considerably, as is shown in an extensive numerical study. Furthermore, the results show that also rather simple Machine Learning approaches are able to reach high forecast accuracy. The optimization model does not only lead to more robust solutions, but also to less actual waiting times for the vessels and hence to an enhanced service quality, as can be shown by studying the resulting schedules for real vessel data. Moreover, it turns out that the accuracy of the resulting berthing schedules, measured as the deviation of planned and actually realisable schedules, exceeds the accuracy of all forecasts which underlines the usefulness of the DTB approach

##### 2020

[182412] |

Title: Robust berth scheduling using machine learning for vessel arrival time prediction. |

in: <em>Flexible Services and Manufacturing Journal</em>. 9 (2022). |

Volume: <strong>35</strong>. Number: |

on pages: 29-69 |

Chapter: |

Editor: |

Publisher: |

Series: |

Address: |

Edition: |

ISBN: |

how published: |

Organization: |

School: |

Institution: |

Type: |

DOI: 10.1007/s10696-022-09462-x |

URL: https://doi.org/10.1007/s10696-022-09462-x |

ARXIVID: |

PMID: |

**Note: **i3lab

**Abstract: **In this work, the potentials of data-driven optimization for the well-known berth allocation problem are studied. The aim of robust berth scheduling is to derive conflict-free vessel assignments at the quay of a terminal, taking into account uncertainty regarding the actual vessel arrival times which may result from external influences as, e.g., cross wind and sea current. In order to achieve robustness, four different Machine Learning methods-from linear regression to an artificial neural network-are employed for vessel arrival time prediction in this work. The different Machine Learning methods are analysed and evaluated with respect to their forecast quality. The calculation and use of so-called dynamic time buffers (DTBs), which are derived from the different AIS-based forecasts and whose length depends on the estimated forecast reliability, in the berth scheduling model enhance the robustness of the resulting schedules considerably, as is shown in an extensive numerical study. Furthermore, the results show that also rather simple Machine Learning approaches are able to reach high forecast accuracy. The optimization model does not only lead to more robust solutions, but also to less actual waiting times for the vessels and hence to an enhanced service quality, as can be shown by studying the resulting schedules for real vessel data. Moreover, it turns out that the accuracy of the resulting berthing schedules, measured as the deviation of planned and actually realisable schedules, exceeds the accuracy of all forecasts which underlines the usefulness of the DTB approach

##### 2019

[182412] |

Title: Robust berth scheduling using machine learning for vessel arrival time prediction. |

in: <em>Flexible Services and Manufacturing Journal</em>. 9 (2022). |

Volume: <strong>35</strong>. Number: |

on pages: 29-69 |

Chapter: |

Editor: |

Publisher: |

Series: |

Address: |

Edition: |

ISBN: |

how published: |

Organization: |

School: |

Institution: |

Type: |

DOI: 10.1007/s10696-022-09462-x |

URL: https://doi.org/10.1007/s10696-022-09462-x |

ARXIVID: |

PMID: |

**Note: **i3lab

**Abstract: **In this work, the potentials of data-driven optimization for the well-known berth allocation problem are studied. The aim of robust berth scheduling is to derive conflict-free vessel assignments at the quay of a terminal, taking into account uncertainty regarding the actual vessel arrival times which may result from external influences as, e.g., cross wind and sea current. In order to achieve robustness, four different Machine Learning methods-from linear regression to an artificial neural network-are employed for vessel arrival time prediction in this work. The different Machine Learning methods are analysed and evaluated with respect to their forecast quality. The calculation and use of so-called dynamic time buffers (DTBs), which are derived from the different AIS-based forecasts and whose length depends on the estimated forecast reliability, in the berth scheduling model enhance the robustness of the resulting schedules considerably, as is shown in an extensive numerical study. Furthermore, the results show that also rather simple Machine Learning approaches are able to reach high forecast accuracy. The optimization model does not only lead to more robust solutions, but also to less actual waiting times for the vessels and hence to an enhanced service quality, as can be shown by studying the resulting schedules for real vessel data. Moreover, it turns out that the accuracy of the resulting berthing schedules, measured as the deviation of planned and actually realisable schedules, exceeds the accuracy of all forecasts which underlines the usefulness of the DTB approach