[Newbies] Squeak for robot imaging and scientific applications

[Robert F. Scheer]

I am using Squeak to control a robot, specifically a RoboMagellan
outdoor navigating robot.  Pretty soon it will be important to confront
how to implement the computer vision on this robot.  The OS is Linux.

One way to do this is to use an external vision system such as RoboRealm
running under MS Windows and communicate between the Squeak controller
and the vision system with sockets.  I'll probably do something like
that but it's only a short-term solution.  RoboRealm is a proprietary,
closed-source application that won't be able to do something I'll want
to do, perhaps something like figuring out how far the robot travels
between video frames by comparing the movement of "interest points".  At
that point I will want some type of vision or imaging library directly
accessible to Squeak.

Python has an imaging library called Python Imaging Library.  This is
one of the strongest justifications some people use to write their
entire robot code in Python in the first place.

Likewise, Python has the SciPy library, which provides a large set of
tools for signal and image processing, as well as linear algebra,
optimization, statistics and so forth.

For example, suppose I wanted to write a Kalman or other probabilistic
filter for "fusing" inputs from a cameras, inertial sensors, wheel
encoders and so forth.  In Python, the matrix operations for doing the
math would be simple calls.  

How would you approach these imaging and quasi-scientific issues in
Squeak?  Would you always end up using external packages and making
interfaces to them through sockets or through special primitives in a
custom VM?

Thanks for your thoughts.

- Robert