Incom ist die Kommunikations-Plattform der Fachhochschule Potsdam

In seiner Funktionalität auf die Lehre in gestalterischen Studiengängen zugeschnitten... Schnittstelle für die moderne Lehre

Incom ist die Kommunikations-Plattform der Fachhochschule Potsdam mehr erfahren

Physical Computing // FIREFLY

enstandem im Kurs Physical Interaction Design bei Reto Wettach. Winter 08/09

TASK.

based on the choosen topic to think about lamps, and ways to turn them on and off, i felt the need, to create the illusion of a living light inprisioned in a glass cube.

OBJECTIVES.

inspired by the movement and apperance of fireflies, the light would have to move/fly freely within the box, and would have to to be able to reach every point within this container.

in order to create the illusion of a living beeing, it would have to react to it‘s surrounding.

SOLUTION.

jUHe0dRe.jpgjUHe0dRe.jpg

movement.

for movement, i build an robotic arm with 3 servo motors.

one for 360° turning (x-axes) one for 180° moving up and down (y-axes), and one for pushing in and out (z-axes).

the whole setup is controlled via arduino and some algorithmns inspired by ben fry's integrator class, allowing a smooth but also firefly typical jumpy movement.

also to get a rid of the cable-mess, i printed a pcb for the basic wireing.

.

closeclose
upup
the difference bettween breadboard...the difference bettween breadboard...
...and pcb...and pcb
the final setupthe final setup
closercloser

light.

to find the right light source, was one of the hardest things.

_the angle of a basic led was to spotty, i wanted a nice diffuse to all directions spreading light source _a normal light bulb was to heavy. _a very small light bulb was still to heavy. _the tiny halogen-lamp, got to hot, and was still too heavy

_finally i found very very tiny led „chip“. wired 5 of them together, one for each side, which gave me the desired result.

flying.

to hide the robotic arm and make only the light visible, a cube with diffused foil was build. „1/2 white diffusion“ had the right transparency.

..

onon
offoff
the boxthe box

senses

to have the light react to it's outside word, i searched all kinds of analoge inputs... piezos,light-, ir- and ultrasonic-sensors.. but none could, other then getting very expensive, detect movement in a quite big radius and also be precise aubout the destination of the movement...

so i took a webcam and stuck an fisheye lens in front of it. it hangs over the lamps and basically detect the whole whole room.

it's all wired to a laptop which does some opencv blob detections via processing and give the coordinates via a serial protocol to my arduino board, which then according to the speed and position of the movement, triggers one of the three algorithmns:

SLEEP (if nothing happens) the fly gets slower and darker for about 1:30 min until it is almost off.

INTERESTED( if the movement is slow and smooth) it jumpy gets near the detected movement, „looks“ around but always stays, in the area of detected movement.

PANIC (if the movement is fast, or „suprising“) very fast jumps around the cube, mainly staying as far as possible away from the movement.

i don't want to go deeper here in all the x,y to degree translations, and programming topics. but so much more to say, that the programm is written, so that some parameters like calibrating, brightness, contrast (of the camera) x rotation adjustments (for the exact alignment of camera and device) are controlable live via a gui, which made the handling very much easier.

for the nerds, geeks and freaks, the source code is listed in the material folder.

84yZeuX4.jpg84yZeuX4.jpg

the webcam+fisheye

COCLUSION.

there are still some problems with the tracking, since, when it is dark, the webcam results get very dirty. but i also could imagine wide angle ir-sensors on each side, which for now, was mainly a cost issue. then also the hooked up notebook wouldn't be nessecery anymore.

the mechanics need to be improved. the toothweels now beeing out of plastic, have to be steel ones, to have a reliable machine, since it breaks too fast, due to the hard pressures of flight impact.

a very positive result is, that the light, somehow, actually seems to have it's own life. there is an interaction and feedback, but never quite predictable like playing with a fly. which i'd never expect to work that good, and made me quite happy, to have created a low-fi artificial intelligence, glowing and inprisoned in a box.

thanx for reading till the bitter end!

Ein Projekt von

Fachgruppe

Sonstiges

Art des Projekts

Studienarbeit im ersten Studienabschnitt

Betreuung

foto: Prof. Reto Wettach foto: Stefan Hermann

Entstehungszeitraum

SoSe 08 – WiSe 08 / 09

zusätzliches Material