IDEA : WWW bike route server
Radek Aster
system will allow a user to determine the "best" bike route between
two street intersections, where best might be (a) traffic density, (b)
directness, (c) ??
The heart of this system will be an undirected graph, with nodes
representing street intersections and arcs representing streets,
expwys., etc., weigthed by distance, traffic density, etc.
A shortest-path algorithm will be used to determine the "best" route
-- this part is fairly easy and well understood.
The difficult part, at least for me, is developing the data base for
the graph. Does anyone know where I might get city/urban/local maps in
machine readable form ? I can't even begin to imagine sitting there
entering a AAA map by hand, where each intersection and street would
have to be entered, though I suppose a tablet and speach recognition
system would help.
Any ideas ?
--
-------------------------------------------------------------------------------
Radek Aster @ | "If humans are so evolved,
mailto:ras...@sgi.com ___~*: _ \ _ | how come they don't have
mailto:ras...@netcom.com '/|~~\ (*)-^+-(*) | opposable thumbs on their
| feet ?"
http://reality.sgi.com/employees/raster/ |
-------------------------------------------------------------------------------
Bill Bushnell
Aster) wrote:
> I've been toying with the idea of writing a WWW bike route server. The
> system will allow a user to determine the "best" bike route between
> two street intersections, where best might be (a) traffic density, (b)
> directness, (c) ??
>
> The heart of this system will be an undirected graph, with nodes
> representing street intersections and arcs representing streets,
> expwys., etc., weigthed by distance, traffic density, etc.
>
> A shortest-path algorithm will be used to determine the "best" route
> -- this part is fairly easy and well understood.
>
> The difficult part, at least for me, is developing the data base for
> the graph. Does anyone know where I might get city/urban/local maps in
> machine readable form ? I can't even begin to imagine sitting there
> entering a AAA map by hand, where each intersection and street would
> have to be entered, though I suppose a tablet and speach recognition
> system would help.
>
> Any ideas ?
Over the past few years I've put together a bike map of the San Francisco
Bay Area such as you describe. The data are currently stored in a MacDraw
Pro file because I want to be able to print the entire "map" and carry it
with me so I can make route decisions in the field.
My map contains nodes which represent intersections of major streets or
points of interest. Nodes have one attribute: altitude. If you were to
store the map as a database, you'd probably want to store latitude and
longitude. Nodes are conncected by arcs whose attributes are distance and
gross climbing in each direction. In database format you would only need
to store gross climbing in one direction. You might also want to store
the names of the roads over which the arc passes, the condition of the
road, and the traffic density. I wouldn't try to store every intersection
of every road because you'll be working at it the rest of your life. It's
amazing how much data can be put on a street map! I usually pick an
intersection that is common or near to many different biking routes, and I
try to keep nodes at least 1-2 miles apart. In most cases, an arc
represents an obvious road or a unique set of roads/paths. If there's a
choice, either the distance/climbing is very close or I put a label next
to the arc indicating the roads over which the data was gathered.
A database would be nice because you could query it. For example one
might want to find a cycle (loop) that passes through nodes A, B, and C,
is at least 50 miles long and has less than 3000 feet of gross climbing.
Once the map is in database format, one could write software that
generates the route graph that can be overlaid on a standard map.
Right now I'm still gathering data (by riding the route on my bike and
taking notes from a calibrated Avocet 50 computer), mostly in the East Bay
and Marin County. An older version of the map is available at:
http://www-leland.stanford.edu/group/sbc
I'll be sending Francis Rotella an updated copy of the map as soon as I'm
finished adding a new chunk of data. You have my permission to copy the
data from my map to construct a database provided that you do not
personally profit from it and provided that whatever software/database
results from your effort be shared freely with the Internet community.
--
Bill Bushnell <bush...@ieor.berkeley.edu>
Daniel Connelly
> I've been toying with the idea of writing a WWW bike route server. The
> system will allow a user to determine the "best" bike route between
> two street intersections, where best might be (a) traffic density, (b)
> directness, (c) ??
>
> A shortest-path algorithm will be used to determine the "best" route
Bill Bushnell replied:
>Over the past few years I've put together a bike map of the San Francisco
>Bay Area such as you describe.
>
>A database would be nice because you could query it. For example one
>might want to find a cycle (loop) that passes through nodes A, B, and C,
>is at least 50 miles long and has less than 3000 feet of gross climbing.
>Once the map is in database format, one could write software that
>generates the route graph that can be overlaid on a standard map.
=============
Map overlays, route-optimization, etc, are all interesting ideas.
However, I question their usefulness relative to the amount of effort
is required for their implementation. Humans are pretty good at
picking out good paths to connect different nodes, and coming up with
a metric for the cost associated with a given route is always going to
be time-dependent and subjective -- mileage alone doesn't do it.
However, a simple utility that generated a text-form route sheet,
along with numbers for net distance, net climbing, and net descending
for a given set of paths from Bill's data (with options of entering
paths not already in the data set) would be quite valuable and
would no doubt get considerable use. It would allow quick generation
of route sheets of the form used in the Sequoia rides.
Dan
Radek Aster
>
> Radak wrote:
^^^^^
Arggg .... RadEk ... but what the heck, it's Monday, spend Sunday
at the JG memorial, so who's counting. :-)
> (with options of entering
> paths not already in the data set) would be quite valuable and
> would no doubt get considerable use.
Excellent idea!!! That sounds like a wonderfull compromise to creating
a big honkin' huge database which, for the most part, will be unused.
By allowing user entry, interesting routes will get done.
--Radek
Dick Blaine
attended a meeting on standardizing symbols.
Many cities have maps showing bike lanes, route and paths and LEVEL OF
DIFFICULTY/EXPERIENCE for that route. The city maps generally show more
detail within the cities and the county map tends to show more interconnect
information. You can contact the traffic engineer/public works department of
each city to obtain these maps. Contact the county marketing dept.
A digital verison of these routes would be a good starting point for you,
whatever approach you decide on.
Dick
Dick Blaine
advertise 12 million streets. I have KEYMAP by a company called SoftKey.
This runs $30 or so. I think that the data base of streets is the same on
these and only the display/mapping software varies.
Dick