Show simple item record

dc.contributor.advisorBatalden, Bjørn-Morten
dc.contributor.authorWang, Yufei
dc.date.accessioned2025-02-03T15:28:23Z
dc.date.available2025-02-03T15:28:23Z
dc.date.embargoEndDate2030-02-18
dc.date.issued2025-02-18
dc.description.abstractWith the introduction of autonomous and remotely-controlled ships, close vessel encounter can be associated with higher risk of collision or near-miss scenarios. To avoid these possible incidents, a higher level of situation awareness is required to support the respective decision-making process in maritime navigation. Advanced Ship Predictors (ASPs) are thus proposed to be a solution framework to enhance situation awareness for ship navigation as the main contribution of this thesis. The research presented in this thesis focuses on local-scale predictions conducted by the ASP, with the typical prediction horizons ranging from 10 to 90 seconds. The workflow of the local-scale ASP is divided into two main parts: vessel navigation state estimation and pivot point (PP)-based trajectory prediction. Kalman filter (KF)-based algorithms, combined with kinematic motion models, are used to estimate vessel navigation states. These estimated states are then employed to determine the PP of the vessel. The predicted trajectory is generated by leveraging the role of the PP. Given that the understanding and applications of the pivot point are widely adopted in maritime navigation education and training, the local-scale prediction is designed to integrate this understanding to the vessel trajectory prediction. The evaluation of the designed local-scale ASP begins with simulated maneuvers conducted in the UiT bridge simulator. It is then followed by sea trials on the UiT research vessel, Ymir RV. The performance of the local-scale ASP has been gradually improved through modifications after each evaluation. The evaluation based on simulated maneuvering data shows that the 90-second position prediction yields a median L2-norm error between 10m and 15m, while the median heading error at the 90th second ranges from 0 to -5 degrees. During sea trials conducted by the Ymir RV, the predictions are calculated over a 10-second horizon. After the vessel entered a steady state following the execution of a new rudder order, the position prediction revealed a maximum L2-norm error of approximately 8.7 meters, while the maximum heading error was about 13 degrees.en_US
dc.description.abstractMed innføringen av autonome og fjernstyrte skip kan nærkontakt mellom fartøy være forbundet med høyere risiko for kollisjoner eller nestenulykker. For å unngå mulige hendelser, som kollisjoner og nestenulykker, kreves et høyere nivå av situasjonsforståelse for å støtte beslutningsprosessen i skipsnavigasjon. Advanced Ship Predictors (ASPs) foreslås derfor som et løsningsrammeverk for å forbedre situasjonsforståelsen for skipsnavigasjon, som hovedbidraget i denne avhandlingen. Forskningen som presenteres i denne avhandlingen fokuserer på lokale prediksjoner utført av ASP, med typiske prediksjonshorisonter som varierer fra 10 til 90 sekunder. Arbeidsflyten for den lokale ASP-en er delt inn i to hoveddeler: estimering av fartøynavigasjonsstater og pivotpunkt (PP)-basert baneprediksjon. Kalman filter (KF)-baserte algoritmer, kombinert med kinematiske bevegelsesmodeller, brukes til å estimere fartøynavigasjonsstater. Disse estimerte tilstandene brukes deretter til å bestemme fartøyets PP. Den predikerte banen genereres ved å utnytte rollen til PP. Gitt at forståelsen og anvendelsen av pivotpunktet er allment akseptert i maritim navigasjonsutdanning og opplæring, er lokal prediksjon designet for å integrere denne forståelsen i fartøyets baneprediksjon. Evalueringen av den utformede lokale ASP-en begynner med simulerte manøvrer utført i UiT-simulator. Deretter følges det opp med sjøprøver på UiTs forskningsfartøy, Ymir RV. Ytelsen til den lokale ASP-en har gradvis blitt forbedret gjennom modifikasjoner etter hver evaluering. Evalueringen basert på simulerte manøverdata viser at 90-sekunders posisjonsforutsigelse gir en median L2-norm feil mellom 10m og 15m, mens median kursfeil etter 90 sekunder varierer fra 0 til -5 grader. Under sjøprøvene utført av Ymir RV ble prediksjonene beregnet over en 10-sekunders horisont. Etter at fartøyet hadde gått inn i en stabil tilstand etter utførelsen av en ny rorordre, avslørte posisjonsforutsigelsen en maksimal L2-norm feil på omtrent 8,7 meter, mens maksimal kursfeil var rundt 13 grader.en_US
dc.description.doctoraltypeph.d.en_US
dc.description.popularabstractInterest in autonomous and remotely-controlled ships is growing as AI proves its capabilities across various industries. However, the introduction of these new types of vessels increases the risk of collisions and near misses during close encounters with other ships. To ensure safer ship navigation, innovative technologies are needed to support decision-making during ship operations. This thesis focuses on the local-scale Advanced Ship Predictors (ASP), which is designed to predict a ship maneuvering behaviors over short periods. The local-scale ASP operates in two main stages. First, the ship's navigation state is estimated using the combination of kinematic motion models and Kalman filter algorithms. These estimates are then used to calculate the ship's pivot point, which helps predict its position and direction. The pivot point is a key concept in navigation, familiar to trained navigators, and plays a crucial role in understanding and predicting ship behavior. The local-scale ASP was tested using both simulated maneuvers from the UiT bridge simulator and data gathered from sea trials conducted with the UiT research vessel, Ymir. The performance of the local-scale ASP has been gradually improved through modifications after each evaluation. In simulated tests, the local-scale ASP predicted the ship's position with an accuracy of 10 to 15 meters over 90 seconds, and direction changes within a 0 to 5-degree error range. During sea trials, the system made 10-second predictions, achieving a position error of around 8.7 meters and a direction error of about 13 degrees after a rudder adjustment. These predictions can provide navigators with early warnings of potential collisions or near misses, providing more time to make crucial decisions. These predictions can provide navigators with early warnings of potential collisions or near misses, providing more time to make crucial decisions. The local-scale ASP also has the potential to assist digital navigators that are anticipated to be developed by AI.en_US
dc.description.sponsorshipUiT internal fund; MARKOM II project.en_US
dc.identifier.isbn978-82-8236-608-3 (trykk), 978-82-8236-609-0 (pdf)
dc.identifier.urihttps://hdl.handle.net/10037/36402
dc.language.isoengen_US
dc.publisherUiT Norges arktiske universiteten_US
dc.publisherUiT The Arctic University of Norwayen_US
dc.relation.haspart<p>Paper I: Wang, Y., Perera, L.P. & Batalden, B.-M. (2022). The Comparison of Two Kinematic Motion Models for Autonomous Shipping Maneuvers. <i>Proceedings of the ASME 2022 41st International Conference on Ocean, Offshore and Arctic Engineering. Volume 5A: Ocean Engineering</i>. Hamburg, Germany, June 5-12, 2022. Published version not available in Munin due to publisher’s restrictions. Published version available at <a href=https://doi.org/10.1115/omae2022-79583>https://doi.org/10.1115/omae2022-79583</a>. <p>Paper II: Wang, Y., Perera, L.P. & Batalden, B.-M. (2023). Kinematic motion models based vessel state estimation to support advanced ship predictors. <i>Ocean Engineering, 286</i>(1), 115503. Also available in Munin at <a href=https://hdl.handle.net/10037/30189>https://hdl.handle.net/10037/30189</a>. <p>Paper III: Wang, Y., Perera, L.P. & Batalden, B.-M. (2023). Coordinate Conversion and Switching Correction to Reduce Vessel Heading-Related Errors in High-Latitude Navigation. <i>The 22nd World Congress of the International Federation of Automatic Control (IFAC2023), 56</i>(2),11602–11607. Yokohama, Japan, July 9-14, 2023. Also available in Munin at <a href=https://hdl.handle.net/10037/32046>https://hdl.handle.net/10037/32046</a>. <p>Paper IV: Wang, Y., Perera, L.P. & Batalden, B.-M. (2024). Adaptive Kalman Filter-Based Estimator with Sea Trail Data to Calculate Ship States in Complex Navigation Conditions. <i>The 34th International Ocean and Polar Engineering Conference (ISOPE2024)</i>, ISOPE-I-24-536. Rhodes, Greece, June 9-14, 2024. (Accepted manuscript version). Published version not available in Munin due to publisher’s restrictions. Published version available at <a href=https://onepetro.org/ISOPEIOPEC/proceedings-abstract/ISOPE24/All-ISOPE24/ISOPE-I-24-536/546234>https://onepetro.org/ISOPEIOPEC/proceedings-abstract/ISOPE24/All-ISOPE24/ISOPE-I-24-536/546234</a>. Accepted manuscript version also available in Munin at <a href=https://hdl.handle.net/10037/35384>https://hdl.handle.net/10037/35384</a>. <p>Paper V: Wang, Y., Perera, L.P. & Batalden, B.-M. (2024). Localized Advanced Ship Predictor for Maritime Situation Awareness with Ship Close Encounter. <i>Ocean Engineering, 306</i>, 117704. Also available in Munin at <a href=https://hdl.handle.net/10037/33502>https://hdl.handle.net/10037/33502</a>. <p>Paper VI: Wang, Y., Perera, L.P. & Batalden, B.-M. (2024). Pivot Point Estimation based Advanced Ship Predictor Evaluation with Vessel Maneuvers under Sea Trial Conditions. (Submitted manuscript).en_US
dc.rights.accessRightsembargoedAccessen_US
dc.rights.holderCopyright 2025 The Author(s)
dc.rights.urihttps://creativecommons.org/licenses/by-nc-sa/4.0en_US
dc.rightsAttribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)en_US
dc.subjectAutonomous shipsen_US
dc.subjectMaritime situation awarenessen_US
dc.subjectShip trajectory predictionen_US
dc.titleLocal-Scale Advanced Ship Predictor towards Enhanced Maritime Situation Awarenessen_US
dc.typeDoctoral thesisen_US
dc.typeDoktorgradsavhandlingen_US


File(s) in this item

Thumbnail
Thumbnail
Thumbnail

This item appears in the following collection(s)

Show simple item record

Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)
Except where otherwise noted, this item's license is described as Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)