AIS for collision avoidance - One for the maths gurus?

[xxyyzz]

Is intellectual curiosity not enough?

 

[AliM]

Fireball

You misunderstood. I agree that Pythagorus works well on a flat earth for the sort of distances we are looking at. I was talking about calculating all that stuff from positions, not bearings and distances.

However, if you take your boat's position from GPS - call them N0,E0 (southern or western hemisphere are -ve), and the ship's position, say N1, E1, then the bearing from you to the ship on a flat earth would be arctan ((N1-n0)/(E1-E0)) with a bit of fiddling to get the angle between 0 and 360 degrees. That's not right on a spherical earth - you need an cos(N0) factor somewhere in that fractio - in the numerator, I think.

 

[AntarcticPilot]

I'm probably being dense, but I still don't see how all this business of entering things into spreadsheets is any quicker, easier (or more accurate) than looking at the movement of the blob to see whether it's heading for the centre of the screen.

On the assumption that "If you can keep your head when all about you are losing theirs, then you probably haven't grasped the seriousness of the situation, my son" I'm guessing that I have missed something. Can anyone tell me what, please?

I agree with Tim; transferring information from one system to another - especially where cut and paste is not an option - is extremely error-prone and unreliable. And if the situation is close enough for it to be worth doing, you wouldn't have time! Checking for unchanging bearing is much the most efficient way of seeing if a collision situation is developing; it works equally well on AIS, Radar and Mk1 Eyeball!

It is always worth asking "What assumptions are going into computations like CPA and so on?" And "What data is going in?" The one thing no-one seems to talk about is, "what is the likely error?" - and the answer for distant situation is likely to be "Lots". Headings (especially your own!) are unlikely to be held to a degree (even under autopilot, a degree is doing well; you probably won't be that good steering by hand); bearings are probably measured with similar accuracy. So, errors in the computation of CPA are likely to be more than 65 metres AT BEST at a distance of 1 mile; proportionately more for greater distances. And that is assuming that the angular errors don't exceed 1 degree - a very optimistic assumption. The error is proportional to the angle error, so if the combined errors of bearing and heading are 2 degrees the error goes to 130 m, and so on.

 

[fireball]

AliM - furry muffs! I admit I didn't even bother looking at the spreadsheet because although it is interesting I didn't have time and have a perfectly good AIS engine & chartplotter that gives me CPA ... now - as AP says - I wonder how they calculate their CPA ...

 

[lw395]

Try thinking graphically.
Plot his position relative to your position
Add your velocity vector to his to get relative velocity (a bit like apparent wind)

This shows how little influence your 5kts has in the face of his 20!

 

[Martin_J]

As JohnChampion and TimBartlett say.. the simple answer is that if the line on the NASA display is ahead of you then the other vessel will be. If behind you then it will be behind you. If the line intersects with your position in the middle of the screen then it's going to hit you.

BTW - make sure whoever is at the helm is steering a straight course - put the autohelm on if necessary or get them to concentrate for a while.. and zoom in temporarily just to see which side of you the track line for the other vessel is showing. It's just like radar in that if you keep a straight course (and so does the other vessel) then all the line on screen will be straight.