We are excited to share the publication of latest study from Inno2mare partner RITEH on Pilot Project 3, where innovative methods for short-term and long-term forecasting of personal watercraft trajectories are explored.
This research presents a Bayesian approach for estimating vessel position, using factors such as heading, speed, time intervals, and offsets of latitude and longitude. Additionally, a Markov chain approach is also introduced to enhance the accuracy of trajectory predictions.
Key highlights of the study:
- Data from a cloud-based marine watercraft tracking system, enabling remote vessel control.
- The impact of weather conditions, including wave height and meteorological reports, on watercraft trajectories.
- Innovative methods for estimating trajectory based on longitude and latitude offsets, speed, heading, and time intervals.
- A long-term forecasting window of up to ten seconds achieved by segmenting non-overlapping trajectories.
- Challenges in long-term forecasting lead to the exploration of more advanced machine learning approaches.
- Results demonstrate that using previous predicted values can accumulate errors, and that environmental variables had minimal impact on model improvement—small watercrafts perform well even in unstable sea states, largely driven by their ability to generate and ride waves.
This research not only deepens understanding of watercraft trajectory forecasting but also opens the door to more sophisticated solutions in future developments.
Read more about the study and its findings at the link below:
