Hi everyone. I’m considering developing a new tool for making navigating at sea safer, simpler and more efficient. In my opinion ECDIS is a great tool for basic navigational tasks, but with the technology we have today the potential for developing new functions is huge. I’m curious to hear from more experienced seafarers than me which additions to ECDIS that could be helpful. I.e., are there any navigational functions that you don’t have today but could be useful?
What I have been thinking about is utilizing sensors better; more efficient association of AIS and radar targets, and in addition add cameras to increase tracking classification and accuracy. I’m also thinking camera targets should be associated with radar and AIS targets to avoid clutter.
Other features I have been thinking about is:
Visualization of CPA. Not only showing DCPA and TCPA numbers, but actually showing the graphical CPA positions on the map.
Dynamic navigational advice. Based on current traffic picture, propose the best way forward. If any target enters path, propose best and most efficient collision free maneuver and draw it on the map.
Advanced target trajectory prediction. As far as I understand, today’s systems assumes constant speed and course. I believe better predictions by using past track and historical data would improve collision alarms etc.
Including a camera feed of surroundings, with automatic identification of targets. Camera targets should be associated with radar and AIS targets to reduce clutter. Camera with night vision.
What do people think about these functions? Could some of these improve current navigational systems? And do anyone have experiences with add-on products for improving navigational safety? Hoping to get a discussion about potential improvements for situational awareness on the bridge.
I guess it could, but would make the implementation a lot harder. And I think the use of VR goggles needs to be normalized somewhere else before putting it onto a bridge. I was thinking this as a separate product in form of a map with additional information, or as integration to an ECDIS.
This must have been considered by many previously for sure? Not the exact implementation, but the concept? Innovating on ECDIS is probably slow because of all the regulations and standards that needs to be fulfilled?
Do you think so? Does Raytheon or any other radar provider have any legal responsibility if for some reason an ARPA track is not accurate, or if CPA/DCPA/TCPA is calculated wrong? It will be the captains responsibility either way I would imagine. At least that’s how I’ve been thinking. You get an advice, but the captain is responsible for making the final choice if this is a safe maneuver to make or not.
One ship I was on a radar with excellent automatic acquisition of targets function. Unlike most this radar could pick targets out of sea / rain clutter without acquiring false targets.
If that’s possible than it should be the standard.
I’ve also read about ARPA and AIS fusion. That’d be worthwhile if it was reliable.
The only ship I ever saw it on was when I was a Cadet on the TSES VI, we weren’t “allowed” to use it so I can’t really vouch for its efficacy, I also wasn’t exactly sweating bullets in dense traffic as a Cadet.
Many years later my experience navigating in heavy traffic would suggest that something like this is probably a nice to have but wouldn’t be heavily used.
Dense traffic situations are dynamic and whatever visual path some algorithm is spitting out on a PPI is going to change so much that if you become overly reliant on it you’ll probably get fired or crash or both. It’s also very possible it wouldn’t show a navigable path – in heavy traffic your radar will often have a CPA of 0 or close it to multiple or many contacts and if you try to trial maneuver your way out of trouble, you’ll have a tough watch (TCPA and common sense become crucial).
Navigation really is equal parts art and science, and dense traffic situations are often best appraised visually (given the requisite experience and seamanship).
RE: Liability
On the same page in the Anschutz manual where the Sea Scout is described, there is a blurb that says you’re responsible for your own collision avoidance decisions.
Just as humans 'fuse" sight and sound to maintain situational awareness (hear a sound, turn to have a look) the information from AIS and ARPA would be fused.
For example when there is so-called “target swapping” the Lat/Long signal from AIS might be useful in locating the correct radar signal.
A search for “Radar Image Processing and AIS Target Fusion” will turn up several technical papers.
I think you’re right in this. Even though AIS can be erroneous, it will in many cases help to “correct” a radar offset and help distinguish targets from each other.
OpenCPN has a decent visualization of this that will graphically show CPA for multiple targets.
I’m not sure I like this as the cognitive load goes from “what’s a safe way out” to “is the proposal safe, and if not, what’s a safe way out?”
However, calculating avoidance maneuvers is relatively straightforward, and could be used to overlay a sector on the chart in which the desired CPA is preserved. I can easily use that to assist my own decision-making as it helps to pre-filter the less-safe options.
This can get complicated. Take a look at this example (ignore the text bubbles). Two of the vessels are turning, which is reflected in their AIS course vectors. Do you know at what point they’ll straighten out? I don’t! I also don’t want to make decisions based on historical “usually they turn here” data. On the other hand, “usually they turn here” does indicate places I might not want to be.
These last two suggest a decision support function which can overlay safer or riskier areas on the chart. My general preference with AI is to use it in ways that support the human, rather than attempting to replace the human. Asking the human to review decisions creates greater chance of error (complacency, etc.) compared to including the human in the decision-making process.
I’m not sure I agree that the OpenCPN shows a decent visualization of the CPA, to me it looked rather messy…
Regarding the turning I very much agree with the “when will the turn stop and straighten out”. That would be a major challenge. But many ships are going in the almost exactly same routes many places, so using historic data would be very nice in many occasions imo.
