Steven Kurtz on Sat, 5 Sep 1998 09:20:37 +0200 (MET DST)

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

<nettime> Critical Art Ensemble & Richard Pell: Contestational Robotics

Contestational Robotics
Critical Art Ensemble & Richard Pell

Keywords: robots/contestation/public space/expression management

Part I

Since the modern notion of the public space has been increasingly
recognized as a bourgeois fantasy that was dead on arrival at its
inception in the 19th century, an urgent need has emerged for continuous
development of tactics to reestablish a means of expression and a space of
temporary autonomy within the realm of the social. This problem has
worsened in the latter half of the 20th century since new electronic media
have advanced surveillance capabilities, which in turn are supported by
stronger and increasingly pervasive police mechanisms that now function in
both presence and absence. Indeed, the need to appropriate social space
has decreased in necessity with the rise of nomadic power vectors and with
the disappearance of borders in regard to multinational corporate
political and economic policy construction; however, on the micro level of
everyday life activity, and within the parameters of physical locality,
spatial appropriations and the disruption of mechanisms for extreme
expression management still have value. Each of us at one point or
another, and to varying degrees, has had to face the constraints of
specific social spaces that are so repressive that any act beyond those of
service to normative comportment, the commodity, or any other component of
the status quo is strictly prohibited. Such situations are most common at
the monuments to capital that dot the urban landscape, but they can also
be witnessed in spectacular moments when extreme repression shines through
the screenal mediator as an alibi for democracy and freedom. The finest
example to date in the US was the 1996 presidential election. A protest
area was constructed at the Republican National Convention where
protesters could sign up for 15 minute intervals during which they were
permitted to speak openly. This political joke played on naive activists
had the paradoxical effect of turning the protesters into street corner
kooks screaming from their soapbox about issues with no history or
context, while at the same time reinforcing the illusion that there is
free speech in the public sphere. Certainly, for anyone who was paying
attention enough to see through the thin glaze of capital's "open
society," this ritualized discontent was the funeral for all the myths of
citizenry, public space, or open discourse. To speak of censorship in this
situation or in the many others that could be cited by any reader, is
deeply foolish, when there was no free speech or open discourse to begin
with. What is really being referred to when the charge of censorship is
made is an increase in expression management and spatial fortification
that surpasses the everyday life expectation of repression. Censorship and
self-censorship (internalized censorship) is our environment of locality,
and it is within this realm that contestational robots perform a useful

The Function of Robots

While robots are generally multifunctional and useful for a broad variety
of duties such as rote tasks, high precision activities, telepresent
operations, data collection, and so on, one function above all other is of
greatest interest to the contestational roboticist. That function is the
ability of robots to insinuate themselves into situations that are
mortally dangerous or otherwise hazardous to humans. Take for example
three robots developed at Carnegie Mellon University. The first is a robot
that can be affixed to pipes with asbestos insulation; it will inch its
way down the pipe cutting away the asbestos and safely collecting the
remains at the same time. For a robot, this one is relatively inexpensive
to produce, and could reduce the costs of removing extremely carcinogenic
materials. The second is a robot designed in case of a nuclear accident.
This robot has the capability of cutting into a nuclear containment tank
of a power plant and testing for the degree of core corruption and area
contamination. Once again, this method is certainly preferable to having a
person suit up in protective gear and doing the inspection h/erself.
Finally, an autonomous military vehicle is under development. The reasons
for the development of this vehicle are not publicly discussed, so let's
just imagine for a moment what they might be. What could an autonomous
military vehicle be used for? Let's make the fair and reasonable
assumption that it has direct military application as a tactical vehicle
(it is a humvee after all). It could have scouting capabilities; since the
vision engines of this vehicle are very advanced this possibility seems
likely. At present, the vehicle has no weapons or weapon mounts. Of
course, such an oversight could be easily remedied. If the vehicle was
used as an assault vehicle it would still follow the model set by the
prior two robots. In other words, it could go into a situation unfit for
humans and take action in response to that environment. However, one
element distinguishes the potential assault vehicle from the other two
robots. While the other two are primarily designed for a physical
function, the latter has a social function--the militarization of space by
an intelligent agent. Of modest fortune is the fact that this model can be
inverted. Militarized social space can be appropriated by robots, and
alternative expressions could be insinuated into the space by robotic
simulations of human actions. While autonomous robotic action in
contestational conditions is beyond the reach of the amateur roboticist,
basic telepresent action may not be. 

The Space of Contestational Robots

Like the physical dangers of being irradiated or breathing asbestos, there
are specific social spaces which are too dangerous for those of
contestational consciousness and subversive intent to enter. Even the
tiniest voice of disruption is met by silencing mechanisms that can range
from ejection from the space to arrest and/or violence. For example, being
in or around the grand majority of governmental spaces and displaying any
form of behavior outside the narrow parameters designated for those spaces
will bring a swift response from authorities. Think back to the example of
the convention protest space. Using the designated protest area was the
only possibility, as no protest permits (an oxymoron) were being issued.
Those who attempted to challenge this extensively managed territory were
promptly told to leave or face arrest. These are the hazardous conditions
under which robotic objectors could be useful by allowing agents of
contestation to enter their discourse into public record, while keeping
the agent at a safe distance from the disturbance. (The remotes can work
up to 90 meters; however, the robot has to be kept within the operator's
line of sight.)

Performative Possibilities

What could a robotic objector do in these spaces? We believe that it could
simulate many of the possibilities for human action within fortified
domains. For example: 

Robotic Graffiti Writers. These robots are basically a combination of a
remote control toy car linked with air brushes and some simple chip
technology. When running smoothly, this robot can lay down slogans (much
like a mobile dot matrix printer) at speeds of 15mph. (See part two.)

Robotic Pamphleteers. Simply distributing information in many spaces (such
as malls, airports, etc.) can get a person arrested. These are the spaces
where a robotic delivery system could come in handy--especially if
deployed in flocks. Remember, that people love cute robots (the
anthropomorphic, round-eyed japanamation cute is a recommended aesthetic
for this variety of robot), and are more likely to take literature from a
robot than from most humans. At the same time, the excessively cute
aesthetic can lead to robotnapping. 

Noise Robots. Very cheap to make from existing parts. Particularly
recommended for indoor situations. By just adding a canned fog horn or
siren to a remote toy car one can create a noise bomb that can disrupt
just about any type of small to medium scale proceeding into which it can
be insinuated. 

These are but a few ideas of how relatively simple technologies could be
used for micro disturbances. Given the subversive imagination of Nettime's
constituency it's easy to believe that better ideas and more efficient
ways of creating such robots will soon be on the table. However, it also
has to be kept in mind that robotic objectors are of greater value as
spectacle than they are as militarized resistance. After all, they are
only toybots. Yet these objects of play can demonstrate what public space
could be, and that there are other potentials in any given area beyond the
authoritarian realities that secured space imposes on those within it. 


There is a triple cost to this type of robotic practice. First, it does
require a modest amount of electrical engineering knowledge, and as we all
know, education costs money. Second, it requires access to basic tools,
but a machine shop would be better. Third is the cost of hardware. Robots
are expensive, and there is no getting around it. In the field of robotics
proper, it is barely possible to build a toy for less than 10,000 USD. We
have brought the cost down to between 100 and 1,000 USD, but this could
add up very quickly for a garage tinkerer or for underfunded artists and
activists. It seems safe to assume that a robot will be used more than
once in most cases, but even so, robotic objectors are outside the
parameters for a common, low cost, tactical weapon. To be sure, this
research is in its experimental stages. 


In spite of the fact that contestational robotics is a completely civil
action and poses no danger to anyone, do not expect authority to share
this belief. First, when placed in a militarized area (i.e., any space in
which deep capital is being protected) robots are assumed to be of
military origin. Given this association, it is likely that the robotic
objector will be perceived as a weapon, and treated accordingly. In
conjunction, the builder of the robot is very likely to be treated as
military personnel. Even if the robot is captured and found to be only a
toy, the builder of the robot will be subject to arrest for hard jail-time
crimes, because the military/police were deployed against a militarized
menace. The charges that an activist may face vary in number and wording
from state to state and from country to country, but they all have one
common function. They give police discretionary arrest privileges. Even
though no violent crime is committed, those associated with the state's
perception of attempted violence can be arrested as if a violent crime had
been committed. Laws against "crimes," such as creating a false public
emergency are regularly used in such situations by authoritarian agencies.
These laws are designed specifically to make it easier to arrest political
dissidents and to stifle determined attempts at open discourse. They are
also a way of re-presenting ethical political protest as terrorist action,
and are one of the state's best sleight of hand tricks. This situation is
very much the same as when hackers are called terrorists even though their
only crime is trespassing in an electronic environment where there is no
one to terrorize. Given this extreme and unjust reaction, be sure to
purchase supplies with cash, wear gloves when building robots, use only
common parts and/or materials, remove serial numbers when necessary, and
do not routinely frequent any supplier. Be careful: capital gets very
reactionary when you hack its technology. 

A Note on the Relationship of Amateurism to Contestational Robotics

The amateur has been a scorned figure in post-enlightenment knowledge
management. Specialists and experts are the ones who get the praise. In
this situation, each knowledge specialist hides in h/er own tower, making
occasional encroachments on neighboring territories. In turn these short
range migrations are rebuked as amateur attempts to marshal information
resources that trespassers cannot understand. This attitude is not totally
without merit. Knowledge specializations are very complex and do require
years of study to master. At the same time, dismissing the amateur out of
hand can have a detrimental impact on the practical aspects of applying a
specialization, whether in the material or policy arenas. Amateurs have
the ability to see through the dominant paradigms, are freer to recombine
elements of paradigms thought long dead, and can apply everyday life
experience to their deliberations. One of the most recent examples is the
tremendous job that amateur scientists and health care practitioners did
in shaping policy regarding HIV. Now most experts wouldn't recognize these
people as scientists or healthcare providers, they were just people living
with AIDS, and/or AIDS activists and/or concerned individuals dedicated to
social justice who collectively had an impact on policy construction.
Their expertise came from everyday life experience and amateur study, and
yet this collection of people who rallied in coalitions such as ACT UP had
remarkable vision. In relation to robotics most of us aren't mechanical
science experts, or software or electrical engineers, but we do have the
advantages of being naive visionaries with collective political
experience, the desire to share skills and resources, and the collective
ability to open any desired field of knowledge. Home tinkering is of
necessity in robotics and biotechnology to the same degree we have seen
success in information and communications technology (everything from
simple shareware to ascii culture to hardware recycling). Praise be to the
tinkerers, to the toy makers, and to the amateurs. New versions of
expertise must be constructed. Without tinkerers using models of anarchist
epistemology, contestational robotics will not come to be. 

Part II

How to Build a Robotic Graffiti Writer

This article is the first in a series of robotic objector projects for the
home roboticist/anarchist. This design combines the integrated perception
and autonomous navigation skills of the human dissident with the
efficiency and compact size of a robot specifically adapted to the tactics
and terrain of street actions.  The basic design calls for a roughly
shoe-box-sized trailer to be drawn by a remote controlled vehicle.  The
trailer consists of an array of five spray paint units that are controlled
by a central processor.  The vehicle is navigated into the target area by
its human operator.  At the appropriate time a switch on the controller is
thrown, signaling the start of the "action."  As the vehicle rolls along
the ground, the row of spray cans prints a text message in much the same
way that a dot-matrix printer would.  For example the word 'CAPITALIST'
would be written as: 

***  *  *** *** * * *** * * *** **
* *  *  *   *   * * *   **  *   * *
***  *  * * **  * * *   *   **  **
*    *  * * *   * * *   **  *   * *
*    *  *** *   *** *** * * *** * *

Depending on the nature of the action, the vehicle can either be navigated
to a secluded "safe-zone" or considered a worthy sacrifice in the name of
robotic objection. 

The skills needed to build this robot do not require an engineering
degree, although they do require a reasonable amount of experience in
building circuits, programming micro-controllers (Basic STAMP), and shop
skills/metal working; the project might best be accomplished by a small
group of individuals. 


REMOTE CONTROL CAR [This will be by far the most costly aspect of this
project.  When coupled with the radio controller and essentials such as a
battery charger, the vehicle represents a roughly $500 investment.  What
makes this car exceptional is that it needs to be capable of pulling 3-4
kilograms of additional weight and still maintain a top speed of 10-15
 This generally means a scaled-down version of a 'Monster Truck' i.e.,
multiple engines, etc.  Consult your local RC enthusiast--they love these
sort of specialty problems.  It also must be able to receive three
channels instead of the usual two.]

RADIO CONTROLLER [Any three channel controller will do.]

2 WHEELS [Light weight street wheels from an RC catalog.]

5 INTERMITTENT SOLENOIDS [The surplus variety will be more than adequate
here.  Something in the neighborhood of 24v (.25 - .3 amp) that can hold
itself shut against fairly vigorous tugging.]

BATTERIES [One to power the solenoids (probably 24v) and one to power the
circuitry (9v).]

5 SPRAY CANS [The 3 oz miniature variety is best for reasons of weight and
size.  However, the industrial paint that road workers use could be used
if the weight is less of a problem.  Remember to choose a color that
complements the terrain.]

MICRO-CONTROLLER [Almost any standard chip (i.e., BASIC stamp) will
suffice as long as it has at least two inputs and five outputs.]

LED/OPTO-TRANSISTOR [for use as an encoder.]

TRANSISTORS, RESISTORS, CAPACITORS, and WIRE [Specific values cannot be
given here, as there are too many variables to worry about.]

RAW MATERIALS [1/32" aluminum or plastic sheet, lightweight plastic or
wood square stock (1/4" by 1/4").]


There are too many variables at work here to describe the construction or
components in extreme detail.  Availability of surplus goods and access to
means of production will vary from group to group. 

As with any robotics project, the strategy is to work on individual parts
AND the overall product AT THE SAME TIME.  One needs to be building
working sub-systems, while continually evaluating them to ensure that they
will work together. 

The project is divided into four subsystems. 

1) Micro Controller (+software)

2) Encoder

3) Structure of Trailer

4) Solenoid->Spray-can system

The Micro Controller: 

A plethora of micro-controllers exist that are easy to use and learn.  Any
of the more popular packages that clutter the pages of 'hobbyist'
magazines will suffice as long as they meet the requirements of having at
least two inputs and five outputs.  The first input pin is used for the
signal that comes from the controller and tells the micro-processor to
start performing its task i.e., print the text.  The second input pin is
for the encoder that attaches to one of the wheels or axles.  The encoder
tells the processor how fast the vehicle is moving in terms of 'clicks'
(see encoder section). Each 'click,' or 1/4 turn of the wheels, will mean
that one column of a letter is to be printed.  This allows the processor
to adjust the space of the letters according to how fast the car is
moving.  The five output pins are all used for controlling the solenoids
that activate the spray cans. 

The Text

As mentioned earlier, the text is printed as if by a dot-matrix printer. 
Each individual letter is printed with a 5 by 3 grid of dots and therefore
requires a minimum of 15 bits to be rendered.  The most cost effective
method of storing this data in terms of RAM would be to use 16 bit blocks
(type SHORT) for each letter in your array and simply ignore the last bit. 
However, if you have the RAM, it may be more elegant to use one byte for
each column (three columns per letter).  This abstracts things a bit,
making it easier to print simple graphics instead of text or to use the
extra bits in each column as a kind of control character.  For instance,
you could have a bit that controls how long the can sprays, making it
possible to have dots and dashes. 

Depending on how much RAM the micro-controller has, you could build a
function into the chip that translates the text into a binary stream using
a lookup table--for instance, 111111010011100 for the letter P, as in the
example earlier.  Such a table would use only around 52 bytes or so (2
bytes per letter times 26 letters).  Or translation could be done offline
and the stream hard-coded into the chip at programming time. 

The following is some pseudo-code that should give a fair idea of how the
components interact with each other.

Typedef COLUMN = a byte

pin1 = GO signal
pin2 = wheel encoder
pin3-7 = solenoids

COLUMN the_text_array[# of letters] = convert_text("THE MESSAGE TO PRINT")

   if(GO signal ON)      //If it gets the GO signal, the loop
      timer + 1          //must run 5 times with the signal ON
   if(GO signal OFF)     //before it will GO.  This prevents false signals
      timer = 0            
   if(timer > 5){
      for(i = 1 to # of letters){
         for(j = 1 to 3){           //The number of columns in a letter
            col = read_next_column(the_text_array)      
            paint_column(col)       //writes the bits to pins 3 thru 7
            wait (for encoder click)
         all pins OFF                  //puts a space between letters
         wait (for encoder click)

Signal from Controller: (i.e., GO!)

The average remote control car uses a minimum of two channels in order to
be controlled by the remote.  That is, one channel controls forward and
backward motion, and the other controls left and right motion.  It is very
easy to add channels by using standard parts from an RC hobbyist catalog. 
In this case, we need one more channel that will be used to trigger the
text printing function.  The signal that comes out of the receiver on the
car is most likely going to be PWM (Pulse Width Mod), in which case the
supplied code should be sufficient to direct the signal straight into the
micro-controller.  Should the signal happen to be analog, most
micro-controllers have at least one pin that can receive an analog signal. 


There's no need to run out and buy a 600-degree optical encoder for this. 
All we need is a standard LED and phototransistor pairing. They tend to
look like this: 
     __   __
     |L| |P|
     | |_| |

There are two standard ways of implementing these as an encoder.  In one
version, the principle works like thus: When the LED light hits the
phototransistor, it is ON.  When something is stuck in between them, it is
OFF.  All we do is attach a pinwheel divided at 45 degree intervals to the
axle of one of the wheels and have it pass through the center of the
pairing, like this: 

Fig 1.
   \  |  /|           |
    \ | / |           |
   __\|/__|           | <- pinwheel
  |  /|\              |
  | / | \          __ | __
  |/  |__\         |L|||P|
                   | |_| |
  pinwheel         |_____|

This is where the 'clicks', described earlier, originate.  Each space in
the pinwheel causes one click in the phototransistor.  The signal from the
transistor is then passed on to pin2 of the micro controller. 

In another variation on the same theme, the LED/phototransistor pair are
pointed at a black and white pinwheel (potentially the wheel hub). The
light from the LED reflects off the white parts and triggers the
phototransistor, sending it into an ON state.  The light is absorbed by
the black sections, sending it into an OFF state. 

Trailer Construction: 

Anything more than a cursory description would be impossible here without
the use of mechanical drawings or photographs (see upcoming web version). 
The basic idea is that we have a trailer chassis resting on two wheels. It
is connected to the rear of the vehicle via some type of flexible joint. 
The chassis can be made out of a sheet of lightweight plastic or aluminum
with plastic or aluminum supports. The spray cans are secured, lying flat
on the trailer between the wheels.  A slot or window runs the width of the
trailer below the spray nozzles and perpendicular to the spray cans (this
is what they spray through).  The solenoids are mounted on a shelf raised
an inch or so above the spray nozzles.  This allows room for the batteries
and electronics to be stored underneath.  (See Fig. 2)

Solenoid--spray-can mechanism: 
Mechanically speaking, this portion will be the most difficult to
construct and will require a lot of kludging to get it right.  What we've
got is a row of five spray-cans facing downward and another row of five
solenoids that must use their 'pulling' motion to 'push' the buttons of
the spray cans.  This is probably most easily achieved by a simple system
of fixed-pivot linkages.  The solenoids are arranged so that they are
facing (plungers toward) the spray nozzles, and probably raised an inch or
so above the nozzle center.  The linkages should in the form of the letter
Z, with joints at the corners and a fixed-pivot point somewhere in the Z
diagonal.  The plungers of the solenoids should attached to the upper
portion of the Z and the lower one will touch the tip of the spray can. 

Fig. 2 (Side View)
      |     |
      | Sol.|=[-------O-joint
 _____|_____|__       |
  ___________ |       o-pivot
  |         | |       |         _________
  |Batteries| |       |        /         |
  |         | | joint-O---- []=|  spray  |
  |_________| |                \_________|
 _____________|____________    ____________

The placement of the pivot point on the linkage determines how much
leverage is placed on the nozzle.  
This may take some tweaking to get enough pressure to make it spray on


The intentions of this article are two-fold.  First, it presents one
concrete example of how a robotic objector can be built to be useful to
resistant forces.  Second, it should open up critical discussion of the
value, implications, and design of these tools.  Several prototypes are
already in the construction phase of development and collective discourse
can only enhance the process. 

#  distributed via nettime-l : no commercial use without permission
#  <nettime> is a closed moderated mailinglist for net criticism,
#  collaborative text filtering and cultural politics of the nets
#  more info: and "info nettime-l" in the msg body
#  URL:  contact: