Small Changes, Big Refactoring: Precision in Voyage Planning
Published on March 17, 2026 by Viktor Kolbasov
It has been a month since our last update, and what was planned as a small change led to a big refactoring. We shifted our focus: it is no longer just about speed, but aimed firmly at precision. In our industry, even missing a small dot on a chart can mean aiming a vessel towards danger, and we definitely don't want that to happen.
Building on Proven Foundations We demonstrated that our code can efficiently handle the conversion of S-57 ENC to structured data using GeoPackage and Postgis storage solutions. In Update v0.1.1, we showed that full workflows can be processed locally on consumer devices in a matter of hours. This effectively solves two of shipping's biggest concerns: internet traffic and security. By enabling an air-gapped system, we comply with shipboard Cyber Security best practices, which is crucial given that most ECDIS updates are still carried out through USB media.
The Current Update:
Reliability and AI-Readiness Our current update builds on these established best practices to solve probably the most crucial part of the workflow: reliably ingesting data from ENC to Graph. We are also future-proofing this process for ML acceleration. The AI revolution is already here and will not go away easily, so we are choosing to ride the wave “responsibly”. This update is already 60-70% ready and will be our first to come with full test coverage.
Reliability and AI-Readiness Our current update builds on these established best practices to solve probably the most crucial part of the workflow: reliably ingesting data from ENC to Graph. We are also future-proofing this process for ML acceleration. The AI revolution is already here and will not go away easily, so we are choosing to ride the wave “responsibly”. This update is already 60-70% ready and will be our first to come with full test coverage.
The Triple-Backend Approach To ensure absolute accuracy, weight assignment will be covered by three backends of choice: Postgis, SQLite + GeoPandas, and a no-SQL approach. Here is why this matters:
- Error Baselining: It gives us a solid baseline for errors; if one backend deviates more than 2% from the others, it raises a red flag.
- Fixing Industry Norms: We have all experienced multimillion-dollar software populating routes with the wrong depth or other data, forcing us to manually prove it wrong. We are addressing that exact experience from the start.
- Built-in Redundancy: We always have a fallback option with minimal impact. If the Postgis server goes down or SQLite has module conflicts, the ship still needs to sail. Our task is to make sure of that.
Looking Ahead:
UI, Trials, and SMM'26 This will probably be our last update before departing for the next voyage. However, because all processing operations will be in order, and thanks to the release of QGIS 4.0, we can finally create a skin for our code. This will add visualization capabilities and allow for graphical input, elevating functionality to a whole new level. Sadly, due to recent hiring issues affecting both time and funding, we are uncertain if we can carry out Pilot trials this year. Instead, we are looking to test our market fit against all market leaders during SMM’26.
UI, Trials, and SMM'26 This will probably be our last update before departing for the next voyage. However, because all processing operations will be in order, and thanks to the release of QGIS 4.0, we can finally create a skin for our code. This will add visualization capabilities and allow for graphical input, elevating functionality to a whole new level. Sadly, due to recent hiring issues affecting both time and funding, we are uncertain if we can carry out Pilot trials this year. Instead, we are looking to test our market fit against all market leaders during SMM’26.
A Call to Our Fellow Seamen Right now, this remains primarily a research and testing tool as an extension of QGIS, but the vision is much greater. At the end of the day, time is the most valuable thing we have. The future of this project depends on ensuring that the time we spend developing it actually saves time for others, especially our fellow seamen. I know how frustrating it is to do your best when your tools are dull and unsharpened, and suggestions to management are met with ignorance. It is time to create the performance we need ourselves.
- If you see a use case for this project, let us know.
- If your voyage planning tools are missing features, let us know.
- If your workflow has highly time-consuming tasks, let us know.
Speak up. Let your voice be heard. We are listening.