David Overholt

Robby the Racist Robot

This installation creates a light-hearted atmosphere for people to consider statistics, stereotyping, racial profiling, and computer/human interaction by taking a single skin-tone scan of the user and creating a statistically founded, yet consistently incorrect and presumptuous profile.


Introduction to Computational Media - Thur (Rozin)

As information, data, and statistics continue to grow as consumables it is highly important for people to understand and pay attention to its use and misuse. This project asks for people to give the system a single source of information (skin tone) which is taken to create a fact-based stereotype (which will always be incorrect). Using algorithms for color proximity, randomization in statistics, and various other methods, the user realizes although a single source, and viewpoint, of information may be relevant and can be used to distinguish between people; characteristics of people and their social environment are much more important.

Study of pigmentation and chromatic scales (Von Luschan), Monte Carlo method to determine random-within-random numbers, Java based application support for multiple XLS documents, serial and parallel dot matrix printing capabilities, shell command scripting, video color tracking, array organization and sorting, color differentiation

Anyone with skin pigmentation. This project is intended to provoke interaction, fun, and consideration between people around the installation.

User Scenario
The user walks up to the device and places their hand into the scanner. When the hand is entered a light fills the area and the scan is taken. The information is then processed and filtered in the computer which speaks and prints the stereotype on an Okidata ML395 dot matrix printer for the user to take with them.

An example of the output may be:

Country of origin: Niger
Sex: Male
Age: 54
Literacy: False
Oil Consumption: 4.5 mL
Expected lifespan: 58.2 years
Internet access: True
Occupation: Agriculture


An upside-down overhead projector case has been turned into the scanning device which includes a proximity sensor, camera and super bright LEDs for consistent lighting. An algorithm then averages the skin colors received and finds its closest relation.

A country of origin is chosen based on this color and information is pulled from ~50 databases of information from the CIA Factbook (among others).

The computer then uses a shell script command to speak the information (age, sex, country of origin, literacy, oil consumption, etc) and prints to an Okidata ML395 dot matrix printer, which is then available for the user to take with them.

Conclusions (stereotypes) will obviously come up different for each person and even a single person will receive very different results from each use.

Color scanners didn't work, camera color tracking needed, broke lots of stuff (like the projector), hard to find dot matrix printer, hard to interface a 20 year old dot matrix printer with a new Macbook Pro and Arduino/Processing