Collision Avoidance, Cones of Uncertainty, and Appropriate CPA

[Seaworthy Lass]

Originally Posted by evm1024

Just as a point of reference:

The stanchion method is referenced to the boat (which does change its heading), The HBC is referenced to magnetic North which is constant tor the time scale we are talking about.

Of course everything you said is true but it is important to understand the uncertainties in out reference points.

Yes, I forgot to add yawing (let alone deliberate changes in heading) and changes in speed, importantly not just ours, but also the other vessel's.
Unlike the stanchion method, the HBC method is unaffected by any of these alterations.

Also, for our purpose, unlike the stanchion method, the HBC method is unaffected by any change in where we are located on the boat while taking measurements (our boats are small and the measuring distances are several nm).

SWL

[Seaworthy Lass]

PS The HBC method is extremely useful, yet I get the impression it is used very little and if used, it is only when a risk of collision has been determined by the stanchion method. You are then starting afresh with a new technique trying to determine at a much closer distance if a risk of collision actually exists.

Electronics seem to be relied on almost exclusively nowadays and I think their accuracy is likely being overestimated.

The discussion regarding "cones of uncertainty" is a valuable one.

SWL

[Dockhead]

Originally Posted by Seaworthy Lass

PS The HBC method is extremely useful, yet I get the impression it is used very little and if used, it is only when a risk of collision has been determined by the stanchion method. You are then starting afresh with a new technique trying to determine at a much closer distance if a risk of collision actually exists.

Electronics seem to be relied on almost exclusively nowadays and I think their accuracy is likely being overestimated.

The discussion regarding "cones of uncertainty" is a valuable one.

SWL

AIS is really a "killer app" for this purpose, giving you near instantaneous calculations of CPA and TCPA, and giving you data on his course and speed which is not even calculated, but communicated to you by him.

So it's kind of natural to focus on this data in trying to identify collision risks -- it is effortless and usually (almost always) far more accurate than data produced in any other way.

BUT as you say, it is not infallible and like all other data used to make life and death decisions, should be cross-checked against information from other sources, including your eyes.

The other problem with AIS, in my opinion, is that it may degrade orientation, since it does not require any kind of visualization of the target and the target's relative motion. For this, radar has huge advantages, especially when used with the EBL. And of course visual observation is immensely valuable. But radar, used with appropriate skill, is in some ways even more broadly useful than your eyeballs, even in good weather. Our radars have rather poor bearing discrimination, but are very accurate for range, and knowing that you can parse an immense amount of information out of what is displayed on your radar screen. Radar is also an instant cross check for AIS, if your setup displays AIS targets on your radar screen.

The HBC is in some ways obsolescent for collision avoidance because it doesn't tell range at all, and it doesn't tell you what kind of CPA you have. It only tells you if the bearing is changing or not, and with experience you get a feel for how fast the bearing needs to be changing at what approximate range to be passing safely. One thing it DOES tell you which is not always simple to get out of commercial AIS displays, is whether you are passing ahead or behind.

Nevertheless, you can do perfectly adequate collision avoidance with it if you follow a disciplined, systematic procedure like the one described by Uricanejack. With good process, you don't actually need precise data, and you will naturally get adequate CPA's because the HBC cannot distinguish a collision course from a pass a few cables away. The process described by Uricanejack is the polar opposite of the "just charge in and correct as you go along" approach advocated by some others. This kind of approach is absolutely key to good collision avoidance, and so I guess another drawback of AIS is that it is so good, and so much reduces the work load required to figure out a crossing, that it will be tempting a lot of people to skip a lot of necessary process -- just get the necessary information in one glance and make a maneuver, without planning ahead.

I'm not saying anything against AIS -- I agree with whoever said above that it's the best thing since GPS for our safety -- just that like so many other modern conveniences, it should not be used as a crutch to support poor procedure.


One other note, concerning your comment that going from stanchion method to HBC may put you awfully close by the time you have finally developed some good data:

One fundamental part of good collision avoidance procedure, is doing the right things at the right time. An extremely common failure of recreational sailors is not having a sufficiently wide horizon of awareness, not understanding the different COLREGs phases of maneuvering, and so doing the wrong things at the wrong time -- standing on, for example, when it was long ago time to maneuver. Someone, I think it was in this very thread, said something like "if I determine at half a mile that a risk of collision exists . . . " -- well, half a mile is far, far too late to be determining whether or not a risk of collision exists or not. When dealing with fast ships in open water, you need to identify risky targets at no less than 8 to 10 miles out(commercial mariners generally use 10 miles for the "identify" phase of a crossing). At that distance, the ship will still be hull down on the horizon. That's the time for the stanchion test, and not later. If you are not using electronics, it takes some minutes (as we saw in Uricanejack's excellent description of good procedure with a HBC) to start to get a clue -- the ship will travel a few miles meanwhile. So if you start doing that much closer than 10 miles out, you may really not have time to identify a problem, make a plan, and then do a maneuver, before you are uncomfortably or even dangerously close.

From Cockcroft, paraphrased here: Ajish Gopalakrishnan: 4 Stages of Collision


"1. At long range, before risk of collision exists, both vessels are free to take any action.

2. When risk of collision first begins to apply the give-way vessel is required to take early and substantial action to achieve a safe passing distance and the other vessel must keep her course and speed.

3. When it becomes apparent that the give-way vessel is not taking appropriate action in compliance with the Rules the stand-on vessel is required to give the whistle signal prescribed in Rule 34(d) and is permitted to take action to avoid collision by her manoeuvre alone, but a power-driven vessel must not alter course to port to avoid another power-driven vessel crossing from her own port side. The give-way vessel is not relieved of her obligation to keep out of the way.

4. When collision cannot be avoided by the give-way vessel alone the stand-on vessel is required to take such action as will best aid to avoid collision.
The distance at which the various stages begin to apply will vary considerably. They will be much greater for high speed vessels involved in a fine head on or fine crossing situation. For a crossing situation involving two power-driven vessels in the open sea it is suggested that the outer limit of
the second stage might be of the order of 5 to 8 miles and that the outer limit for the third stage would be about 2 to 3 miles.

It is essential for good collision avoidance that both vessels have similar views about when these phases occur -- you can't just make it up as you go along.

[terah]

Originally Posted by Dockhead

It is essential for good collision avoidance that both vessels have similar views about when these phases occur -- you can't just make it up as you go along.

That's one of the (to me) tricky bits of collision avoidance, as it changes so much depending on circumstance. Stage 3 when racing round the cans can easily be seconds before a collision. In the 5kt vs 20kt crossing from this thread there's clearly a wide difference of opinion (was there ever a consensus?). The ship and the yacht are never going to have the same view of the phases unless they consider them from the other's perspective.

The other one that gets me is what happens when your view of the crossing changes. This could be because you realise a mistake, or because the crossing itself has changed. I had this last summer in Greece with a day tripper boat. It was on my port bow, heading towards me at about 15kt, and looking to cross my bow with a CPA of maybe 200ft, but then he tightened up to more like 50ft. I suspect he was quite in control of the crossing from his point of view, but from mine it was getting tight so I made a significant turn to port.

[Lodesman]

Originally Posted by Seaworthy Lass

PS The HBC method is extremely useful, yet I get the impression it is used very little and if used, it is only when a risk of collision has been determined by the stanchion method.

Electronics seem to be relied on almost exclusively nowadays and I think their accuracy is likely being overestimated.

Professionally, I take gyro bearings (gyro-compass based on true north) of any vessel that comes into view. When starting out, I jotted the figures down on a tote, but with time I was able to keep track of several mentally. This was always step one, as it also gave me a bearing to check on radar. Then I would check the radar, and plot or hook the paint. I don't use AIS for anti-collision, although I will check the speed and course reported by AIS as a check of my radar or visual calculations. I like AIS for putting names to the contacts.

On my own boat, it's much the same - primary tool is HBC, and I check every vessel that comes into view, other than those obvious situations where risk of collision doesn't exist (eg. red to red aspects). If I have the radar up, I use the MARPA, as limited as it is. I again use AIS to doublecheck my assessments, and get the names.

[Dockhead]

Originally Posted by Lodesman

Professionally, I take gyro bearings (gyro-compass based on true north) of any vessel that comes into view. When starting out, I jotted the figures down on a tote, but with time I was able to keep track of several mentally. This was always step one, as it also gave me a bearing to check on radar. Then I would check the radar, and plot or hook the paint. I don't use AIS for anti-collision, although I will check the speed and course reported by AIS as a check of my radar or visual calculations. I like AIS for putting names to the contacts.

On my own boat, it's much the same - primary tool is HBC, and I check every vessel that comes into view, other than those obvious situations where risk of collision doesn't exist (eg. red to red aspects). If I have the radar up, I use the MARPA, as limited as it is. I again use AIS to doublecheck my assessments, and get the names.

Interesting that you would use MARPA before AIS. Is it habit? Or do you think there is some inherent advantage? I'm interested that I hear over and over again from pros like you that they only use AIS as a kind of secondary method. But even if it were working well, MARPA takes longer to produce a plot, takes longer to respond to course/speed changes, and is much less precise than AIS, isn't it?

MARPA has never worked adequately on any of my radars, so I don't often use it. I do like the view of relative motion on the radar screen, however, and I often use the EBL.

[Uricanejack]

Originally Posted by Dockhead

Interesting that you would use MARPA before AIS. Is it habit? Or do you think there is some inherent advantage? I'm interested that I hear over and over again from pros like you that they only use AIS as a kind of secondary method. But even if it were working well, MARPA takes longer to produce a plot, takes longer to respond to course/speed changes, and is much less precise than AIS, isn't it?

MARPA has never worked adequately on any of my radars, so I don't often use it. I do like the view of relative motion on the radar screen, however, and I often use the EBL.

Ha ha

Of course its a habit.
A good habit, based on knowledge, experience and generally accepted good practice. The electronics are always secondary information used to improve upon not replace the primary observation information.

Just because an instrument appears to give very accurate information. It does not mean it is accurate information.

[Lodesman]

Originally Posted by Dockhead

Interesting that you would use MARPA before AIS. Is it habit? Or do you think there is some inherent advantage? I'm interested that I hear over and over again from pros like you that they only use AIS as a kind of secondary method. But even if it were working well, MARPA takes longer to produce a plot, takes longer to respond to course/speed changes, and is much less precise than AIS, isn't it?

Haven't really given it much thought before; I guess some of it might be 'habit' - started out that way before AIS existed, and am comfortable with it. Some of it might be that I have not been overly impressed by AIS. Even with Cl A units, I'm surprised by the number of ships that just disappear or pop up out of nowhere, and some never display at all. Given the slow update rate of my Cl B, I wouldn't put much stock in the data, as it was only valid in that msec that the update sent. I have more faith in MARPA as it is, because I can generally compensate for the variability of my own course and speed; the relative lag in updates does 'smooth' the bumps in the data, and I have faith in the data I'm using - not reliant on the other ship sending accurate info.

[DotDun]

Originally Posted by Dockhead

Interesting that you would use MARPA before AIS. Is it habit? Or do you think there is some inherent advantage? I'm interested that I hear over and over again from pros like you that they only use AIS as a kind of secondary method. But even if it were working well, MARPA takes longer to produce a plot, takes longer to respond to course/speed changes, and is much less precise than AIS, isn't it?

MARPA has never worked adequately on any of my radars, so I don't often use it. I do like the view of relative motion on the radar screen, however, and I often use the EBL.

FWIW, here are some shots of ARPA vs AIS on my boat. Furuno ARPA has much faster updates than Class B AIS. I don't remember what caused the divergence between the two in the last few shots, but guessing by the returns to the right, it may have been wave action rocking my boat.

[svrevelations]

This thread has been very informative - thank you.

I want to create a calculating collision avoidance Acrobat Reader file, when certain values are added, the form will do mathematical calculations and provide detailed information. Whilst this form may not be practical whilst at sea on a collision course, but it can serve for exercises and become acquainted with time, distance and CPA.

Kindly look at the attached Acrobat Reader file, I would appreciate your help with the math formulas required to do the calculations. I did PM Dockhead in this regard, but he has not responded.

The input fields are marked A, B, C, D and E. I need the formulas for 1, 2, 3, 4 and 5. In Acrobat, the calculations have to be written in Java, it will simply things if you have any knowledge of this. If not, an Excel spreadsheet will suffice and I will try and get the formulas converted to Java.

If I can get it done;

• I will make the Acrobat Reader file available to this forum,
• I will give you due credit and on the Acrobat Reader form.

[transmitterdan]

For closest point of approach formulas look at the source code of OpenCPN. It computes CPA with data about two vessels position, course over ground and speed over ground.

[wjhutchings]

For those of us without the electronic gizmos take a series bearings of the vessel if the bearing stays steady then you are on collision course, if it is closing then it will pass ahead of you, if the bearing opens then it will pass astern of you, if you are end on or nearly end on then if you see either of the side lights on their own then they will pass clear down the side if you both side lights then you are on collision course.

[jackdale]

Originally Posted by DotDun

FWIW, here are some shots of ARPA vs AIS on my boat. Furuno ARPA has much faster updates than Class B AIS. I don't remember what caused the divergence between the two in the last few shots, but guessing by the returns to the right, it may have been wave action rocking my boat.

A class A AIS updates every 2-10 seconds.

Position updates for Class B transponders are broadcast less often than Class A transponders. Vessels going less than 2 knots transmit position updates every 3 minutes while vessels traveling more than 2 knots transmit position information every 30 seconds. (https://www.milltechmarine.com/faq.htm#a6)

That might account for the difference.

[Lodesman]

Originally Posted by wjhutchings

For those of us without the electronic gizmos take a series bearings of the vessel if the bearing stays steady then you are on collision course, if it is closing then it will pass ahead of you, if the bearing opens then it will pass astern of you, if you are end on or nearly end on then if you see either of the side lights on their own then they will pass clear down the side if you both side lights then you are on collision course.

Bearings are not normally described as "closing" or "opening" - that would be how one refers to range. The bearing could be said to increase/decrease, draw right/left, or veer/back. If it's on your port side and the bearing is decreasing, then it will pass astern of you; if on your starboard side, the bearing would need to increase for it to pass astern of you.

[Jim Cate]

Originally Posted by jackdale

A class A AIS updates every 2-10 seconds.

Position updates for Class B transponders are broadcast less often than Class A transponders. Vessels going less than 2 knots transmit position updates every 3 minutes while vessels traveling more than 2 knots transmit position information every 30 seconds. (https://www.milltechmarine.com/faq.htm#a6)

That might account for the difference.

From the point of view of the yacht, it does not matter how often a B class AIS updates the data. It is how often the ships AIS updates, and the ship will likely have A class AIS. From the ship's pov, the slower update rate, coupled with the relatively erratic speed and course of the yacht might lead to inaccurate CPA calculations, and hence influence their choice between ARPA and AIS for decision making.

Jim