ian.baldwin

Another great piece of software if you need to edit/view meshes is Meshlab:

The only limitation is the maximum number of vertices that are saveable into the 3ds format.

This code demonstrates the MOOS codebase operating with the Location Manager in iOS.

Download the source code

Get the project on github.

A current project required a small diagnostic utility to run at boot-time to report on the status of various modules and sensors, and I found myself writing a utility to check the status of various FireWire cameras in python, which led naturally to the absolutely fantastic pydc1394 project. Using pydc1394 in my code, I found that I could write an image-logging device driver in all of 48 lines of code:

This simple driver logged at frame-rate for several hours, with two different types of PointGrey cameras with minimal modification.

FFMPEG is a great swiss-army tool for video manipulation. I recently had need to use ffmpeg and ffserver to pass a video feed – constructed of static images – to a remote server, for wider redistribution. Searching google, the top hit for “piping rgb data to ffmpeg” results in this article from Kyle Cordes, which is several years old. In the article, Kyle details his efforts in piping raw data into ffmpeg, and the requisite conversion from RGB to the yuv4mpegpipe format.

There has apparently been quite some progress on ffmpeg since that article, as now piping data into ffmpeg is relatively easy. Firstly, create your pipe:

$ mkfifo image_pipe

Then, using your own code, write the data to the pipe in RGB tuples:

...
fifo f();
f.open( "image_pipe");
std::vector<image>::iterator it = images.begin();
while( it != images.end() )
    f.write( it->m_data, it->height, it->width );
...

Then, invoke ffmpeg with the command:

$ ffmpeg -f rawvideo -pix_fmt rgb24 -s 512x384 -y -an -i \
 ./image_pipe http://localhost:8090/image_feed.ffm

which will stream the images direct to your server for redistribution.

Using the Motion Strategy Library

The Motion Strategy Library (MSL) developed by Steve LaValle @Illinois is a fantastic stochastic-planning library. With some of the dependencies of the library having recently changed, have a gander at what you need to do to get the code up and running.

Here are some questions I have been receiving about compiling MSL:

• Q.Configure complains about not finding libFOX
• A. Ensure that you have installed the FOX libraries in a place that configure knows about, or, use the –with-fox-libs directive.
• Q.Of course I have done that, but it still can’t find the libraries
• A. Ensure that you have re-named libFOX-1.X.XX.a to libFOX.a or add in a symbolic link.
• Q.Great, now how do I divorce the planning engine from the GUI?
• A. There is a gui-free example in the tests folder named - nogui.c Using the code in this way requires you to have some way of interfacing your geometry with that of the planner.