Open Source Navigation

[MaineCub]

Guys;

Thank you all for your input. There definitely are alot of things that need to be taken into consideration. I definitely have ALOT to learn about Navigation, GPS, GIS and such.

I think my initial goal is to start small, get some tools built. Then we can all determine HOW the tools work, the parameters etc, ie: Business Rules.

A journey of 1000 miles always begins with the first step.. Ooh wait.. With the first sail!!

Best,

--Dave

[coot]

Originally Posted by DeepFrz

I don't know how people got the idea that Dave has been talking about chart making. This is about writing Open Source software to be used for navigational purposes, not creating charts. The data in this instance is "code", not chart data points. We were talking about CVS/Subversion for a software code depository and setting up an IDE and data base system to get started. I guess it's nice to know how charts were plotted in the old days if one were interested in that sort of thing.

Deep.

I think the idea came from these sentences in the very first post of this thread:

But I don't see how difficult it would be to 'tie' a GPS to a depth finder and a recorder of some sorts. The recorder could then on a timely basis report it's data back to a central location. Subscribers could download the data into their navigation machines etc and use.

What I see here is:
- gps and depth finder data recorded -> a list of depths and positions
- report data back to a central location -> put it on a chart
- download data to navigation machine -> distribute the chart

If you don't think the resulting software is going to make charts, then what is it going to do?

[Pblais]

If you don't think the resulting software is going to make charts, then what is it going to to?

It will make a disjointed bathymetric model with no reference points. The data will be "fuzzy" at best when you account for the depth and location errors.

I worked on a project that did this on land with from a moving vehicle. The real process for doing this type of work is a lot more complex if you want data that you could really use. Old style charts are far more accurate than this proposed method would yield. The more data you collected the worse it would get.

As far as an Open Source project goes the you can find CVS forums that can help you in ways this forum can't. It's not like we don't allow it but you could get all the information from them better than you can here.

If you want a good product to look at from a point of view of how you can do these types of projects check out the Paint.NET project. It's actually a first class graphics tool much like Paint Shop Pro version 9.0. It's also a free product.

Using something like the new WinFX product and .NET could get this product going a lot faster plus result in something worth using.The trouble with most all the free packages available is they really are not very easy to use, don't do much and support only the oldest chart formats.

[DeepFrz]

I can see where the difference of opinion comes in here. I took it to mean depth info, maybe hazards, etc. I think plotting these onto electronic charts is different than producing charts. Maybe not.

Paul, if someone reported the DGPS position of a rock (for instance) that wasn't charted or was charted in a "wrong" position, wouldn't that be valid information?

Also, if a number of reports of depth info through say a passage came in and there was a program to sift through this information looking for congruence(?), wouldn't that be valid information?

Deep

[Pblais]

Paul, if someone reported the DGPS position of a rock (for instance) that wasn't charted or was charted in a "wrong" position, wouldn't that be valid information?

Anything could be valid but perhaps not accurate. Uncorrected GPS is not accurate enough for chart quality. Real charts are actually that good in terms of the base they are plotted on. True DGPS is corrected from a local base station. The idea is you measure the GPS from a static location and then use the GPS time stamps to add the correction to the recorded data point. The GPS signal has a highly accurate time signal and it is actually why GPS can be so accurate. Consider GPS signal to be a very very accurate clock. You really can take the data saved from a base station and match it with the data point you collected in the field and compute the error at the base station and apply it to the field data point. Through yet another process you can record data from a static location for a very long period of time and eventually compute the location to some amazing accuracies. It's how they measure the movement of the continental plates.

US charts are actually more accurate than you might think. They start with low elevation raitioed and rectified air photos then apply them to a control grid.

So if I give you a GPS point of one particular spot there is the error of the raw data I collected combined with the error of your chart, combined with the error of your own GPS when you try to find it. So the rock you located might be wrong because the base is worng or the location is wrong but you can't know which.

That is just one point on the ground. As you add points you add the issue of relative error. If you measure two spots then compute the distance between them the margin of error doubles relative to each other. That is why all charts and all land maps have a control grid that is developed from a highly precise process. It prevents error "creep" and allows you to tie back to a control point to restrict the error. As you get into large area maps like sea charts the whole shape of the earth gets to make this process even more complicated. Above I ignore all that and for local charts you pretty much assume a flat earth. For larger scale charts you can't or you get charts that look like those old maps from the 16th century.

Coastal Charts are made pretty much the same way they make USGS topo maps and are about equal quality. Converting them to electronic makes them no more nor no less accurate. In fact all the NOAA charts are made electronically and have been for many years. The format they use is a Maptech format not used for consumer charts. It's designed as an editor format.

The problem with "reported information" is mostly one of accuracy. Say you have a channel into a port. Of course they silt and shift with the storms. We all know of locations like that. Just reporting simple GPS/Depth won't be accurate enough to correct the chart to the same level of accuracy it was before. The relative GPS error won't locate accurately enough and the depth data has so much variation and problems I wouldn't guess how inaccurate that data "could be". You always have to take data with a pessimistic outlook. Now it gets even worse. We have our shifting cahnnel out there and we get new data all the time. So just when is the data to be thrown out because it is too old? How would you know? Now what about 10,000 ports.

Watching local channels you navigate and observing them over time is what amounts to "Local Knowlwedge".

To rechart a small area you need to collect the data all at one time. You need to collect it such that it defines all the boundaries not just the centerline. You need the depths of where you can't go as well as where you can go so you can locate the "edge" as accurately as you located the raw depths. When you are navigating you want to know where the edges are as much as just a single centerline. How much sea room is there? You want enough points to describe a 3 dimensional space.

Also, if a number of reports of depth info through say a passage came in and there was a program to sift through this information looking for congruence(?), wouldn't that be valid information?

Not unless it is based in magic. I don't see how it could do that what so ever unless it was done in real time. Depth is a serious issue as a moving boat is actually pretty bad for depth accuracy. The display is "dumbed" down so the numbers don't "jitter" as you look at it. The GPS has limits as well. A couple boat lengths is as good as you could bet money on. It might not be that good after we compute the NMEA data transmission time errors and the averaging that goes with it. When I did this in a vehicle we were using millitary data bus speeds like they use on aircraft instrumentation to capture data direct to disk. We couldn't do it real time since we had to post proces it all with the base station data. We also had a ring laser gyro to compute the roll, pitch, yaw errors too.

The bigger problem is every boat goes down the center. You would need a lot of boats going aground to find the edges of the channel. You only know about what gets reported. That one single problem you can't overcome even if you could build a real time trasnmitter and shore receiver (you could). 100 boats come in / out of the channel all down the middle. What new information did you collect? You measure day after day and slowly the channel is shifting. When do you know it moved? How do you distribute the information? What if it moves back?

For your first job. Fine me a rock that is located wrong and then compute how wrong it is. And fianlly, how did you know it was wrong in the first place? For myself, I consider not hitting the rock good enough.

[GordMay]

When I log a waypoint, I might describe it as:
Wpt. #69 - “NW Chanel Marker - 100 yards North of (Eastbound from #68 Russell Bkn @ 101 deg. T)”
25.28.14N x 078.09.63W
Aside from any technical GPS inaccuracies, I’ve also introduced other potential errors:
Was I truly “North” of, or perhaps NNW ?
Was I truly 100 Yards off ?
Was my steering actually 101 degrees true ?
Obviously, my waypoint is NOT accurate to the 6 or 7 significant digits given (implied accuracy of 100 Feet, or about 12 seconds of travel time at 5 Kts).
Good enough for my personal charts, but not accurate enough to publish.

[Sandero]

When navigating around unseen hazards even when "charted" you need to drop the notion of deep on precision and use the notion of margin of error. Until we get to some type of instantaneous real time surveying and updating of our inboard charts... you are olways dealing with historical data.

So you need to take this into consideration. If your draft is say 5'... would you consider 6' depth safe water?.. 8'? How close do you want to play it? Most prudent navigators take all this into consideration and only drop their margin of error when they have no option.

We have run aground in channels marked 7' reported and we have a 6' draft... Looks like that info was old and the channel needs to be dredged. Life and learn and proceed with caution!

Jef
sv Shiva
Contest 36s

[Pblais]

Gord,

Very good example. For my own route way points I never place them on a bouy, but on the deep water side and away from them enough to be sure I'm going there and arrive on the proper side. I'll plan on being at the helm to actuallty arrive there as I might not be exactly correct in my plotting or the chart was wrong and if I'm "close" I can visually verify and with my depth sounder / radar arrive to exactly where I wanted to be.

I say a way point is more like a safe neighborhood you want to be inside so plot the point inside a big safe neigborbood knowing you can't pilot the boat plus or minus a meter from 12 nm out. I assume it it not that accurate and can treat it as such. You use data differently when you know it's not perfect.

There is navigation and then there is piloting. Navigation is the precise calulation. The dealing with imprecise information is what makes the best pilots. In a shallow channel or a tricky spot I want my pilot hat not a GPS. You don't naviagte a tight spot you pilot. There is a lot of other information that is more important than the GPS when you pilot. The world is not accurate enough to navigate every place all the way, but you can't get any place without some navigation. You switch between naviagtor and pilot as required, but it's the piloting skills that tell you when to switch.

For a good exercise plot an electronic course into your slip and see if it can navigate you into the slip. If you get lucky, repeat at different tides and crosswinds. Then do it blinfolded. That is why plotting exact GPS points is never going to get you into the slip.

[DeepFrz]

Jef, good point. Bin there done that. I guess maybe the ability to post a navigation note to a pass/anchorage/area would do just a well or better than a string of depth soundings.

Something like: depth in pass is charted at 8 ft. however it is reported to be less than 6 ft. as of Jan. 01/2006. A thorough survey was not taken so it is not known if the depth accross the entire channel is less than 6 ft. Extreme caution is advised. <GR>

After all, most charted depths are from very old information so any updated info may prove valuable.

Oh, by the way, I don't really mean to imply that hazards etc. would be plotted with "precision", only that there is an uncharted rock at this gps location (as referenced from my gps receiver). This is better than no report, or a chart notation (point) that puts a rocky shelf at one position (ie: charted as a rock, not a shelf or a number of rocks). This happens a lot when information is extrapolated from one chart to another.

Deep.

[Pblais]

Something like: depth in pass is charted at 8 ft. however it is reported to be less than 6 ft. as of Jan. 01/2006. A thorough survey was not taken so it is not known if the depth accross the entire channel is less than 6 ft. Extreme caution is advised. <GR>

If you think about it there is more information in that statement than you could convey with any number of points. The reference to the chart is the common link someone else will use to take your information far more seriously.

Much like what you'll read in any crusing guide. They skip details and tell you things the chart does not show or opportunities you might miss or perhaps a verbal route to an anchorage. Exact precson sometimes is not that useful as you are piloting along trying to find the anchorage. References to the chart are good becuase you probably will have the chart or chart plotter handy too.

The rock you found might better be described in more context even if you also wanted to add the GPS. Visual references help alot. Even something like 100 yards north of the green 3 marker is a great deal of information even if not precise.

[rtbates]

Dave wrote," But I don't see how difficult it would be to 'tie' a GPS to a depth finder and a recorder of some sorts."

Yes it's been done and several manufactures provide fully integrated systems.
The wheel has already been invented. I guess I'm a tad confused by your question. Are you interested in creating a database so DIYers could do the same thing as the professionals? Why? As an engineer who debugs broken code for a living I really can't think of a worse place to have buggy code than in a navigational system. I learned long ago to leave certain things to the experts.

Just my .02

[sinbad7]

Hasn't anyone here heard of the excellent OpenCPN nav software?
Why re-invent the wheel? Reads BSB and VECTOR charts.

Tore

[motorbaad]

Tore not shure they heard about Open CPN in year 2006.

[sinbad7]

OOoooppss...
Back to the future huh?

Tore