Hi,
I was introduced to Processing as a part of a course taught by Dr.Fishwick a couple of months ago. Its an amazing programming software!! I've tried some very basic stuff of simulating some basic 2D Simultaneous localization and mapping (SLAM) with a robot with side-scanning distance sensors (much the popular side-scanning Laser sensors eg. by Sick).

Its in it beta stage and the robot gets stuck at some corners in the room, so in case it does happen you can reset its position with a mouse click. I am working on using the image panning hack by matt.patey to incorporate surveillance of larger floor plans.

I'm a bit fascinated for having implemented a SLAM algorithm(simplistic version of it actually) right now so have put it on my homepage

http://milindshastri.googlepages.com/

Milind

Mapping: The objective is to make the robot traverse through an unknown floor map and be able to generate a map based on the vector V1 + V2; where V1=position vector of robot & V2=position vector of obstacle. To put it very simply, this is the coarse algorithm for generating the map on the right side of the screen.

Obstacles: The obstacles are defined as colored pixels which lie a 3D distance within a threshold from a specified color. In this case Ive defined the obstacle color to be (50,50,50) and the threshold distance as 90.

Sensor: The sensor is a made by rays casts from the robots front up to a fixed range and checking pixel values matching the obstacles color. The rays are drawn from -theta to +theta, in fixed angular intervals, relative to the rotational orientation of the robot.

Avoidance rule: The most essential avoidance rule is
ThetaDirectionRobot+= thetaObstacle/radiusObstacle*30*0.6*random(0.5,1.5);
Acceleration=0.3/(100-radiusObstacle);
Speed+=Acceleration; //max and min speeds are defined so that it doesn't go too fast (front & back)

With this avoidance logic, the robot avoids dark pixels( obstacles) and if it gets close to one it rotates proportional to angular position of obstacle and inversely proporitonal to distance of obstacle and decelerates too (AceFace: This is why it gets out of the bathroom door most of the time). Once it moves away from obstacles within visible range it moves straight and gains speed.

Image Panning: For panning through a large image floor plan I followed http://processing.org/hacks/hacks:offscreen?s=pan%20image . It worked wonders :)

Ive finally got the path finding A* search implemented! The AI library by Aaron Steed http://www.robotacid.com/PBeta/AILibrary/  was very helpful. Thanks.

You can hover your mouse over the grid on a region you want the robot to reach and the robot will find a way, keep updating the obstacles it has seen and continue to calculate the shortest path with the latest available information. Pretty much like a surveillance/ reconnaissance mission  where the user can only see the sensor acquired data (right half of the screen) .

I will cleanup the code and publish a better version soon.
Cheers

Milind

update> I've shifted the applet inside the website. It would need some exploring on http://milindshastri.googlepages.com/projects to find. Sorry for the inconvenience.