A Better Breed of GPS Software?

willo asks: "I recently built an on board computer for my Grand Cherokee. The initial uses for it include music, gps navigation, on board diagnostics and a baby cam so I can see how my kid is in that rear facing seat. After lots of research and testing, I'm really disappointed with the mapping software out there for Linux. Gpsdrive provides the basic functionality I need, but the street names are built into the image and are difficult to read at a glance while driving. Not to mention that it has to download the maps it needs ahead of time. Xastir can handle almost any map out there, but it reads through every map for each redraw! It also seems to lack the ability to zoom intelligently based on location. Note that it's not really designed to be a navigation aid, but rather a ham radio APRS tool. (I am a ham). Delorme Street Atlas USA does what I really want, but it's been a pain to make run properly under wine. Is anyone else out there working on a decent navigation application?"

"To be really usable navigation software should do the following:

• handle maps efficiently and draw them quickly
• have intelligently organized map sets for countries/states. (You can't download a friggin map in the middle of Montana!)
• include serial gps/gpsd support. (just about everyone has this)

I've found a few references to mapping projects that seem to just be vaporware.

Map Sources do exist for this! Bruce Perens made TIGER/line data availible. NAVTECH is the map source for pretty much all the vehicle navigation systems out there, and high resolution maps are availible from the Geographical Information Survey."

Comments

Bring pressure to bear on the vendors

[wowbagger]

I concur that the mapping and trip planning software for any OS:(OS != Windows) is pretty poor. Yes, there are mapping programs out there that will show you where you are, but there are few trip planning programs that will help you select your route, identify items of interest along your route, and help you schedule your time. Of the Windows programs that I've used I have to give the nod to the various Delorme packages. However, Delorme is pretty much Windows only (yes, they do have MacOS versions of some of their programs, but the Mac versions lag the Windows version significantly).

As I see it, the problem is that creating a trip planning program is hard - not only do you need the map data (which is available), you need the route computing algorithm, you need the map drawing algorithm, you need the database of attractions, rest stops, gas stations, restaurants, and hotels. You need to pull all of that together. A proper trip planning program would be a project of greater magnitude than a good word processor.

While I would definitely agree that the best solution would be a Free Software solution, with a Free Software database, if you look closely at my ID you will notice that I'm Wowbagger, not RMS - I will accept a closed solution.

Unless someone is (actually, severel someones are) interested in creating a project on Sourceforce for this, I think the best solution is to bring pressure to bear on the vendors. Imagine if Delorme's suggestion box were /.'ed with requests for a Linux version (or even a Wine version) of their software - they might actually consider it. (In point of truth, setting here on my desk is a mailing from Delorme for their newest Street Atlas program - I am going to reply nicely to it and inform them that until such time as they can tell me truthfully that I can run their program under Linux they can remove me from their mailing list).

I have to wonder if the "give away the razor, sell the blades" model would work for a Delorme - make the actual mapping program Free Software, and make the old versions of their maps freely available (a la the AlladinSoft Ghostscript model), but sell access to the latest maps. This way, we could create a good mapping program under Linux, and have access to a database to feed it, but Delorme could still make money off their value-added - the up to date databases.

What I'd like to have is a scriping interface - so that I could write my planning script to say "Don't bother looking for hotels along the way - only look within 50 miles of a nighttime stop. Look for GOOD attractions within 100 miles of the route, mediocre attractions within 25 miles, and whatever within 5 miles. I want to leave at 17:30 Friday, I want to stop around 21:00 +/- 1 hour, start my days at 07:00, and stop for lunch around 12:00+/- 1 hour. Try to schedule stuff around those times." In other words, let my script generate trial routes, query the database relative to that route, and make changes.

Also, give me the ability to plug in my own databases (something Delorme is lacking in) - I want to get the database of 2 meter and 70 cm repeaters from the ARRL and plug that in too.

Absolutely Correct!

[toupsie]

Thank you! I was about to post a comment along the same lines. The more distractions a driver has in the car, the more dangerous a driver they are.

As my high school football coach/drivers ed teacher used to say, "Keep your God damn eyes on the road before I smack you in the head!".

If it ain't vector, it ain't worth it.

[Andy Dodd]

And from your description, your software is just like gpsdrive, which schleps maps from other sources like MapBlast, Terraserver, etc.

Raster maps don't allow zooming unless you download multiple resolutions, and once you've done this, the maps take a LOT more space than a vector map.

Note: The TIGER datasets are not a good example of a vector map size-wise, as they're in an ASCII format for ease of processing - A lot of the numeric fields could have their ASCII representations replaced with binary to save a LOT of space. Also, rather than having each "road segment" include the name of the road, the segments can be chained together to make a road definition that is a series of latitude/longitude points in a chain. Somewhere I've got a Perl script that does this and outputs it in a format used by a neato Garmin Mapsource map generator program. (Long since disappeared from the 'net... :( )

There's apparently a Gnomad project that aims to support vector maps under Linux, but it's vaporware at the moment. Not even an alpha or CVS release.

The data is there, it's just a matter of writing the software...

Linux, Marketing, and Value

[SEWilco]

Linux? What's that? "
From their email I got a thing like "Market forces.... drop support for Mac.... not in the foreseeable future'

I used to hear that from hardware companies five years ago. I'd invite them to ask their technical staff about Linux, as a significant percent of them are using Linux at home (How did I know? That's what I heard from companies where tech staff answered queries.)...and if Linux has penetrated widely enough that even your staff is using it, consider again its market forces.

I stopped hearing "What's Linux?" from hardware manufacturers within a year. Even their marketing staff now knows.

Small software companies are less likely to know of alternatives, as their few people are less likely to have much experience. (The math is obvious: there are a huge number of small software companies and few people with much experience)

The advantage which many companies have is in their data, not their software. Mapping, geneological, news organizations are selling data.

In the case of mapping, all those dependent upon publicly available data are fighting a losing battle as the public data is massaged and used for public software. Routing algorithms can be a marketing advantage briefly, as there are public routing algorithms and a little public code (some not intended for geographic mapping, such as in circuit layout).

The companies which generate their own mapping data have the greatest data advantage and should be selling the data for any application which wants to use it. Note that any set of data might be sold to a single customer if that customer is willing to pay a sufficiently large price. The creator/user of that data only has an advantage if their data is not available from someone else. And that's a matter of whether or not someone else wants to devote the amount of effort needed to create the data from the publicly available real world. Technology is a strong influence -- mappers used to have to have people extracting data points from photographs, maps, and surveys. Now a computer might be scanning a photo and comparing it with previous maps.

There also are specialty maps, where again it's the effort to create the data which makes it of value. The market for maps of underground mines is small but there are a few mining companies willing to pay a large price. A larger number, such as companies involved in drilling, is less concerned about underground hollows but would be willing to pay a smaller amount for purposes such as reducing the shock on equipment. A much larger number is willing to pay even less, such as prospectors looking for types of minerals previously encountered, environmental groups examining an area, or home shoppers wanting to get a site suitable for trying a geothermal system.

The problem is in how to market the data so as to maximize the profit. Publishing all the data for the many cheap users might mean the few high-priced users won't need to pay the high price. Such things can be handled by methods such as making the price cheaper with a decreasing resolution. The cheapest data might only show there is a mine within a circle of a certain size. The most expensive may be a three-dimensional representation of all the mines in a specific area or highest price for the most recent data.

So... How can you present a business case to Delorme or someone else to make the definition of one of their data sets public? Do they have a product which already only has a data advantage over public code? For years people have been buying CD-ROMs of games solely to get the data file needed for the public code. If a public map program does what I want, I'd gladly buy a reasonably priced set of data for it. Just as I have bought computers preloaded with software priced at thousands of dollars, and immediately erased it all to put Linux on the hardware.