[OSM-dev] GSOC 2009 - Android navigation application

Iulian Banaga iulian.banaga at gmail.com
Wed Apr 1 19:33:44 BST 2009

Hello Marcus,

On Wed, Apr 1, 2009 at 4:05 PM, <marcus.wolschon at googlemail.com> wrote:

> Hello Iulian!
> On Wed, 1 Apr 2009 07:50:19 -0400, Iulian Banaga <iulian.banaga at gmail.com>
> wrote:
> > At that time I thought that one of the best solutions to do all of these
> > tasks is Navit.
> Wasn't Android Java-only or is there support for applications in C?

You can create C/C++ libraries and call them from Android (this is actually
something that I took in consideration but I didn't go further because until
now the performace of the Java VM is satisfactory). What I have done so far
is 100% Java code but it uses a lot of the ideas from Navit (not the actual

> > I studied the application in more detail and I choose o
> > use the same model for data. I had to write my own converter from OSM
> > format
> > to my own binary format.
> What was your motivation to use your own binary format?
> Did you look at OsmBin?
> What was the shortcommint in Navit's format?

As I said my format is based on the Navit's one, and I can say that these
are 80% similar. How did I choose this format? The fact that this app will
run on mobile devices that are limited in terms of memory and CPU I tried
the existing routing applications on my desktop and I was looking for the
fastest renderer and in the same time one renderer that looks good. At that
moment Navit seemed the fastest so I spent some time to understand the ideas
that Navit is using to render.

> > Right now I am starting on the routing part and I am ready to
> incorporate
> > some ideas from Navit. Besides that I believe I will have to use some
> ideas
> > presented in Gosmore and TravelerSalesman.
> Can you elaborate on this some more?
> What ideas do you want to combine?

First of all I was thinking about getting support for routing restriction in
my app, and Nic (Roets) have done a great job with Gosmore about these. From
TravelingSalesman I might use the idea of creating a routing interface that
will be implemented by different algorithms. Both these ideas will probably
be taken into consideration in the second part of GSOC timeline (starting
July) as I would like to concentrate on getting basic routing functionality
until the mid-term evaluation.

> > It will be very nice if I will be accepted for GSOC 2009, as this
> > application will definitely get finished in time as I will sustain my
> > bachelor thesis in late June. I am waiting for any ideas on the routing
> > engine as this is the part that will be most wanted by the end users and
> > also the part that will probably require the most time to be able to get
> > results in a reasonable time on a mobile device.
> Marcus

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20090401/5c12d177/attachment.html>

More information about the dev mailing list