g e n e r a l a r t i c l e
Senster
Reactivation of a Cybernetic Sculpture
A n n A O l S z e w S k A a n d M A R e k D ł u g O S z
T
C
A
R
T
S
B
A
The turbulent history of Senster, a large cybernetic sculpture designed by
Edward Ihnatowicz c. 1970, is divided in two periods: its creation and
prematurely canceled display (1968–1974) and its recent reactivation
(2017–2018). This article presents a comprehensive narrative of
Senster’s reactivation. It explains how the formulation of the conservation
philosophy and methodology employed in the process was instrumental
in delivering the solutions. Based on these observations, the authors
propose a detailed strategy for the maintenance and reactivation of
interactive embodied systems.
What determines the life span of a complex autonomous sys-
tem? We expect that objects of this kind, which some view
as testaments of contemporary culture, will survive into the
future. After a few decades of experimentation with data pro-
cessing and control system engineering, experts in media art
conservation have gradually identified methods that prom-
ise to overcome these obstacles. Museums and art galleries
developed sophisticated procedures to emulate, record or
digitize collected objects. The procedures in question cre-
ate the basis for allographic—inscription-based—presenta-
tions of ephemeral media artworks. As a matter of course,
the inscription-oriented procedures favor specific qualities
of the original item, such as computability, logical coherence
or compliance with current documentation practice, such as
filmmaking or 3D scanning. We discuss below the character-
istics and potential of these qualities in reference to Senster,
a specific example of a hardware-based autonomous system.
Re:SenSteR
Senster merges the concept of kinetic sculpture with the prin-
ciples of cybernetics (Fig. 1). The large-scale work, created by
London-based artist Edward Ihnatowicz, c. 1970, is a classic
early example of media art. However, until recently the piece
was known exclusively via three minutes of footage and a
few archival photos [1]. The work was initially installed at
its commissioner’s newly opened exhibition hall, the Phil-
ips Evoluon, in Eindhoven. Despite its significance, Senster
was dismantled in the mid-1970s as the company decided to
cease its involvement in this costly and demanding endeavor.
Taken out of public view, the piece retreated to the cultural
Fig. 1. Edward Ihnatowicz, Senster, 2.5 m (tripod) × 3 m (arm),
1970, Evoluon Eindhoven. (© James Gardner Archive, University
of Brighton Design Archives)
margins, along with many other icons of a short-lived fasci-
nation with cybernetics.
In 2009, author Olszewska proposed the reactivation
project described here, called Re:Senster, as fellows of the
newly established Faculty of Humanities at AGH University
of Science and Technology in Kraków considered art-related
projects appropriate to a technological academy incorporat-
ing humanities in its curriculum. Re-creation of the historic
autonomous system seemed to converge with the idea of
networking academics, designers and engineers as a multi-
disciplinary team within the academy.
From its inception, the project’s aim was to re-create the
experience evoked by an intertwining of the piece’s form and
movement. The initial plan was to replicate Senster based
Anna Olszewska (researcher, curator), AGH University of Science and
Technology, Faculty of Humanities, Gramatyka 8a, 30-046 Kraków, Poland.
Email: aolsz@agh.edu.pl. ORCID: 0000-0003-3580-4458.
Marek Długosz (researcher), AGH University of Science and Technology, Faculty of
Electrical Engineering, Automatics, Computer Science and Biomedical Engineering,
al. A. Mickiewicza 30 / B130-059 Kraków, Poland. Email: mdlugosz@agh.edu.pl.
ORCID: 0000-0001-6827-9149.
See https://direct.mit.edu/leon/issue/54/3 for supplemental files associated
with this issue.
©2021 ISAST
https://doi.org/10.1162/leon_a_01828
LEONARDO, Vol. 54, No. 3, pp. 299–305, 2021 299
Downloaded from http://direct.mit.edu/leon/article-pdf/54/3/299/1925060/leon_a_01828.pdf by guest on 08 September 2023
Fig. 2. Screenshot of the 3D scan, Colijnsplaat, 2014. (© WH AGH. Image: Marek Ba´scik.)
on archival records and 3D scans the team made after we
located the original in Colijnsplaat (Zeeland, the Nether-
lands) (Fig. 2). However, detailed inspection of the piece in
2017 showed that substantial parts of the skeleton and solid
elements of the mechanical system were stable enough to be
revivified. Despite the absence of control units, damage to
hydraulic pipes, erosion of oil filters and disappearance of
the “head” except for its mounting, the overall structure was
still intact. The quality of the steel and the construction have
made the skeleton resistant to deformation and strain. It is
a mechanical system composed of solid hydraulic pistons
and heavy-duty servo valves, designed according to aircraft
industry and military standards. The research team therefore
decided to keep the remaining original parts of the piece
rather than build a replica from scratch.
In the narrow context of the project’s history, the choice
of a restoration strategy for Senster relied on assessing the
state of the preservation of the sculpture. In the broader
context, there is yet another factor to note. It was possible to
trade radical preservation for functional restoration due to
the project’s siting within an institution. The university esti-
mated the potential risk of intervention as acceptable. This
estimation possibly reflected the stakeholder’s professional
experience and performance-oriented working practices [2].
Hence, the project described here did not follow the model of
retirement, as in the case of Jean Tinguely’s Sculpture méta-
mécanique automobile or Ihnatowicz’s 2013 SAM replication
[3,4]. Instead of displaying the immobilized original next to a
functional replica, we combined and reactivated the remains
of Senster with replicated elements (Figs 3–5).
In terms of contemporary restoration practice, the resto-
ration process reported here corresponds with reactivation
strategies described by Paul Brobbel and Simon Rees in refer-
ence to the restoration of Len Lye’s Loop (1964) and Trilogy
(1977) [5]. In each case, the priority became the approxima-
tion of the piece’s function. Control systems were upgraded
to modern programmable logic controller (PLC) units. Both
Senster’s and Loop’s re-created performance relies on analysis
of the historical footage rather than on the study of control
switches or software.
What characterizes the Re:Senster strategy in the context
of the material obsolescence treatment is that we treated the
traces of wear visible in the structure of Ihnatowicz’s sculp-
ture as aesthetically significant. Hence, we retained scratches,
traces of rust and old layers of paint covering the truss struc-
ture and secured them using an anticorrosion coating. Our
decision to keep some parts of the original mechanics (Fig. 4)
has also influenced the installation’s performance. Due to
minor leakages in the restored pistons and servo valves, the
movements may become less precise, regulated by the con-
trol offset. In order to secure obsolete parts, the program
does not allow the mechanical system to run at full speed.
Interventions into the skeleton were minor. We repaired
some parts of the mechanical system, including three pistons
and four out of six servo valves. We documented and stored
replaced elements. We completed linear position potentiom-
eters, Doppler sensors and the characteristic horn antennas
according to the types used in the original setup. We sub-
stituted other electronic elements with a modern PLC and
microcontrollers (Figs 4–6) [6].
As work progressed, we formulated the maintenance re-
gime for the piece by testing any initial assumptions on the
question of which parts might have been designed for their
artistic merit and which were engineered according to their
function. During the months spent with Senster, we real-
ized that these two complementary design principles would
dictate a different approach and degree of complexity to the
restoration tasks. It was much more challenging to convey
the spirit of the artistic elements, while the restoration of
components such as pistons or filters (verified only gradually
during the project due to previously acquired knowledge of
tested applications of engineering procedures) has proven
300 Olszewska and Długosz, Senster
Downloaded from http://direct.mit.edu/leon/article-pdf/54/3/299/1925060/leon_a_01828.pdf by guest on 08 September 2023
Fig. 3. Senster at AGH, September 2018. (© WH AGH. Photo: Adam Z˙a˛dło.)
Fig. 4. Interventions made during the restoration process; repaired elements (left) and replaced elements (right). (© WH AGH. Photo: Adam Z˙a˛dło. Design: A.O.)
Fig. 5. (left) The original “head,” c. 1970. (© James Gardner Archive, University of Brighton Design Archives.) (right) Its replica, 2018.
(© WH AGH. Photo: Adam Z˙a˛dło.)
Downloaded from http://direct.mit.edu/leon/article-pdf/54/3/299/1925060/leon_a_01828.pdf by guest on 08 September 2023
Olszewska and Długosz, Senster 301
Fig. 6. Senster, detail
of the arm, 2018.
(© WH AGH.
Photo: Adam Z˙a˛dło.)
much quicker and straightforward. Therefore, we based a
restoration methodology on the distinction between those
calculated and freely designed components. The sections
below explain the procedure’s development.
ReACTivATiOn PhASe One:
ReSTORATiOn OF The PhySiCAl COMPOnenTS
Restoration of the physical components began with the
piece’s transportation to Kraków in April 2017 and continued
until October 2018. We treated the skeleton and the mechani-
cal system and then reconstructed the head. Work on the
sensors and wiring concluded the process.
Initially, we viewed a material part of the piece as if it had
been a work of sculpture rather than of calculation and pro-
gramming. Similarly, we thought that only the engineering
principia rather than the artistic values were relevant to the
control and mechanical systems. Our views on the division
between engineered and freely designed elements changed
gradually during the project. Mechanical engineering stan-
dards proved to be key to the restoration of the skeleton.
Grzegorz Biliński and Marek Chołoniewski first argued in
favor of applying these standards as we considered ways to
disassemble the sculpture during the feasibility stage. Dur-
ing subsequent phases of the project, the rules of applied
mechanics provided the principal point of reference for the
mechanical system designers. Jerzy Stojek, Jarosław Mam-
carczyk, Kamil Sikora and Jerzy Hawryluk reverse-calculated
the parameters of missing actuators and servo valves and the
hydraulic pump.
As the construction of the skeleton and mechanical system
followed an engineering blueprint, it was easy to predict in-
terventions into the original structure. For the same reason,
we were able to re-create missing or destroyed parts of the
sensor system. Familiarity with technical specifications of the
original parts ensured that the replacements would conform
in size and proportion with the originals. For example, we
were able to correctly reconstruct the shape of the missing
horn-like antennas attached to the head because their form
was derived from the waveguide resonance frequency of the
original Gunn diodes.
The degree of complexity in the restoration of freely de-
signed components became apparent during re-creation of
the head. Although the mechanical sections of the skeleton
were easy to reconfigure, for those parts where the artist
departed from the clear-cut truss-like structure in favor of
soft molded shapes filled with micro pistons, rebuilding was
more challenging. The pair of vertebrae-like forms moving
independently in XY directions was installed on top of the
arm. They supported a cluster of microphones designed to
detect the presence of viewers. This original part was miss-
ing in 2017. Largely because of this, the reconstruction of the
head by Jacek Żakowski proved to be much more challeng-
ing than the work on the rest of the skeleton. We 3D-printed
three mock-ups of the head before we could agree on its final
shape. We formed the core in the final stages of the proj-
ect, after a trial period that allowed accurate estimation and
capture of the proportions fitting the movement patterns, as
well as the detail and configuration of its mounting (Fig. 7).
Our conclusions regarding the varying degree of complex-
ity in approaching the restoration of elements of a differing
nature are in line with the precepts of Nelson Goodman’s
notation theory. They confirm his famous reflections on the
differences between systems organized as is a musical score,
based on notations, and those systems such as painting,
which he describes as dense, continuous and nonreducible
to a score-like framework of principles. Goodman asked why
people consider a painted image to validly exist only in a sin-
gle, original version, with every other version remaining only
a copy (or a forgery), while with a musical piece, each perfor-
mance can be treated as preserving its authenticity [7]. As we
worked with Senster, we realized that the parts based on the
engineered calculations are comparable to the “score”-based
302 Olszewska and Długosz, Senster
Downloaded from http://direct.mit.edu/leon/article-pdf/54/3/299/1925060/leon_a_01828.pdf by guest on 08 September 2023
Fig. 7. Senster, installation block diagram: electronics (dark), hydraulics (light). (© Anna Olszewska)
pieces. Th ey can be reconstructed without signifi cant loss
of their authenticity. According to the notation theory, ev-
ery new version would be equal, like a musical performance
based on the orchestral score. However, reconstruction of the
artistic elements would be more problematic. Since there is
no score underlying these elements, they manifest in dense,
irreducible and nonreplicable structures similar to those of
a painting. During the restoration we learned that both the
continuous and script-based components are embedded
throughout: starting from the skeleton through the mechani-
cal system, sensors and circuits to the control program.
ReACTivATiOn PhASe TwO:
COnTROl SySTeM DeSign
Th e second phase of restoration focused on the restoration of
movement and interaction functions (June–October 2018).
In this context, we assessed the eff orts in the original control
program implementation to be less signifi cant compared to
the goal of performative feature maintenance. Th erefore, ini-
tially our work with movement functions was informed by
reference to the short footage showing public interplay with
Senster at Evoluon.
Th is phase was particularly demanding due to the dis-
jointed nature of the archival resources. Documentation on
Senster is incomplete, particularly in relation to the analog
electronic components. As a result, there were two options
for the straightforward re-creation of the performance func-
tions. We could have used a version of the original program
compiled by Peter Lundahl and Ihnatowicz in December
1970 [8]. However, this scenario, relying on an archived ver-
sion of assembly code, was potentially less feasible. Th erefore,
the project team—author Długosz, assisted by Rafał Biesz-
czad and Piotr Madej—realized that in designing the new
control system, in order to correctly implement the code, it
was necessary to replicate all the control and signal process-
ing units composing the original system: a missing predictor
that smoothed the movements, an acceleration splitter that
synchronized the speed of the movement and the computer
that processed the subroutines. Physical reconstruction of
these parts would make Senster’s electronics evocative of its
technological history. Th e replication of such a control sys-
tem surpassed the project’s scope and was deferred for future
consideration.
Th e team decided that the original code would be sacri-
fi ced for the benefi t of achieving the original performance
functions. Consequently, the starting point was a formal
analysis of the movement. Th is involved collating fragmen-
tary descriptions of Senster’s behavior with the visual sources.
It was essential to bear in mind that the piece could perform
two modes of movement: tracking and retreat. Ihnatowicz
himself declared that the tracking reaction of the system
would not be proportional to the input signals and “sud-
den movements or loud noise would make it shy away” [9].
Curator James Gardner confi rmed this in his 1988 report,
acknowledging that the tracking would have been performed
up to the point when noise or movements of the viewers
became so intense that it would overload the control system.
Th e curator noted: “As instructions were being shouted at
[Senster] non-stop the computer was stretched to its limits,
and so when the public got too excited, we programmed it
to hold its head in the air—as if to say ‘Enough’ ” [10]. Th us,
Senster’s alarm mode was a safety valve for the relatively slow
data processing system, and the noted “shyness” of Senster’s
behavior was a creative response to the limitations of con-
temporary technology rather than refl ecting any dramatic
intention. By the same token, the artist probably did not de-
cide what level of noise would be suffi ciently high to alarm
the system. Once again, just as in the case of the restoration of
the physical skeleton, enumeration of engineering-calculated
solutions provided a systematic framework for the rest of
the work.
Th ese conclusions are verifi ed by an analysis of short foot-
Downloaded from http://direct.mit.edu/leon/article-pdf/54/3/299/1925060/leon_a_01828.pdf by guest on 08 September 2023
Olszewska and Długosz, Senster 303
age showing Senster interacting with the public on three oc-
casions [11]. The film, when watched in slow motion, exposes
the performance patterns. It shows the arm in constant mo-
tion, alternating between vertical and horizontal. Whenever
a human approached, the preprogrammed path of movement
changed. Initially the arm scanned a wide area in front of the
sculpture and, subsequently, this was limited to just a portion
of the range. The artist’s decision as to how to structure the
sequence of movements probably resulted from the location
of the highest audio signal amplitude detected during each
previous step. Considering the scale of the sculpture (c. 3-m-
long arm), the maximum range of the movement (c. 120°)
and the average space occupied by a human spectator, this
simple method could have led to moving Senster’s head in
front of the viewer within a reasonably short time.
Based on the outcomes of the performance analysis and
the material structure of the piece, we have written a basic
program that could produce a highly simplified interactive
mode. This re-creates the original performance by linking it
to the rudiments of the information theory. We based the pro-
gramming on the assumption that the interactive sequence
observed in the documentary film was characterized by some
degree of redundancy. It should therefore be performed in a
reasonably short, but not the shortest possible, sequence of
movements. The other assumption concerned the external
signals, which were treated as a series of stochastic events
that could engage a value 0 or 1 for every swipe of the arm.
We designed the whole sequence as follows: During the
first stage, the movement covers the whole 120° range in front
of the sculpture. Simultaneously the vertical pair of micro-
phones registers the level of sound amplitude. In this way,
an array of data corresponding to the positions of the arm is
created. This is then quantified into several packets imitat-
ing the data samples, reflecting the slow processing capacity
of the original setup. The comparison of the sampled data
determines the direction of the next movement. With each
subsequent swipe, the range of movement is reduced by half
until the head stops. If the predominant source of sound is
stable, the head should end up in front of the viewer within
three steps.
The second mode was meant to show that Senster is pri-
marily a kinetic sculpture. This mode is more interpretative
then reconstructive. It was designed based on the assumption
that the sinusoid movement pattern is the most common in
nature. For this reason, Długosz proposed that the sculpture’s
arm should move softly along a sinusoidal trajectory with
low frequency. Whenever an external sound impulse was de-
tected, the trajectory of the arm was modified, and Senster
would start to track the sound source. The tracking was based
on sound direction measurements acquired by implementing
the binaural model of soundwave phase measurements. The
time difference between a signal detected by the horizontal
pair of microphones was recalculated into radians and sent
to the main controller.
Once these two complementary modes were showcased,
we found each method to resonate with various display
conditions. The basic interactive program works better in
crowded and noisy surroundings, while the kinetic version
is suitable for a quieter environment with only a few viewers.
Interaction patterns remain open to further development, as
in the case of reactions to movement enabled by the Gunn
transceivers using the Doppler effect (here, we limited the
restoration works exclusively to hardware implementation).
COnCluSiOnS:
SCRiPT TO DeSign MAinTenAnCe STRATegy
Regarding the experience gained during works on the
Re:Senster project, we propose a “script to design” strategy for
restoration of interactive pieces. A meticulous assessment of
both the engineered and freely designed components forms
the core of the feasibility study. This should include all ele-
ments of the piece, including its program, sensors, power
supply system, mechanics and physical parts. Reconstruc-
tion, conservation and repair works should only proceed
based on the knowledge gained through such a holistic as-
sessment. The next stages likewise do not have to follow a
standard skeleton → mechanics → control system progression.
Work with the engineered parts should take precedence over
work on the freely designed components. All reverse engi-
neering should therefore be done during this phase, whether
it relates to the control system, mechanics or construction
engineering. The engineering components are a priority
in dictating the scope and sequence of the restoration proj-
ect and a benchmark for the reconstruction of the freely
designed elements. We expect that various configurations
of these qualities would characterize a broader class of
hardware-based interactive systems.
Senster’s history shows that, despite the ephemerality
of electronic matter, complex cybernetic objects will most
probably function in a historical framework as constantly
evolving entities. The case confirms that the life span of the
complex system depends on qualities such as compliance
with the predominant documentation practice, logical co-
herence and computability. However, these factors do not
guarantee optimal preservation of any such piece. On the
one hand, a hardware-based physical structure such as the
one discussed above cannot be utterly transformed into a
stream of data. On the other, the original codes do not seem
sufficiently culturally valued yet to become the subject of
time-consuming reimplementation that would adapt them
into a renewed structure.
The process described here was limited to a single case
study. One can find comparative material for evaluation of
the proposed strategy in studies on kinetic and media art res-
toration. We can only hope that our conclusions contribute
to the advancement of maintenance methods, with reference
to the issues of sequencing works on the partially preserved
structures. We also hope that readers will find our experience
of interest and accept it as a valid contribution to the ongo-
ing debate on the principal merits of autonomous systems.
304 Olszewska and Długosz, Senster
Downloaded from http://direct.mit.edu/leon/article-pdf/54/3/299/1925060/leon_a_01828.pdf by guest on 08 September 2023
Acknowledgments
The authors wish to express their gratitude to Joanna Walewska for
kindly sharing her knowledge on the location of the archival source
material concerning the history of Edward Ihnatowicz’s piece.
References and Notes
1 For further details concerning the history of the piece, the reacti-
vation project’s origins and members of the team, please see this
article’s online supplementary material.
2 P. Falcão, “Risk Assessment as a Tool in the Conservation of Software-
Based Artworks,” The Electronic Media Review 2 (2011–2012): http://
resources.conservation-us.org/emg-review/volume-two-2011-2012
/falcao (accessed 17 March 2019).
3 R. Bek, “A Question of KinEthics,” in R. Rivenc and R. Bek, eds., Keep
It Moving? Conserving Kinetic Art (Los Angeles: Getty Publications,
2018) pp. 6–15.
4 R. Mulholland, “Play It Again SAM: Replicating Cybernetic Sculp-
ture Using 3D Printing,” V&A Blog (18 November 2013): www.vam
.ac.uk/blog/conservation-blog/play-it-again-sam-replicating-cyber
netic-sculpture-3d-printing (accessed 27 March 2019).
5 P. Brobbel and S. Rees, “ ‘Pretty Good for the 21st Century’: Restora-
tion, Reconstruction, and Realization of Len Lye’s ‘Tangible Motion
Sculpture,’ ” in Rivenc and Bek [3] pp. 120–131.
6 Documentation of the project is preserved in the AGH Faculty of
Humanities archives. Reports on the current state of mechanics, con-
trol system and skeleton are being prepared for publication at www
.senster.agh.edu.pl/reports.
7 N. Goodman, Languages of Art: An Approach to a Theory of Symbols
(Indianapolis: Hackett Publishing, 1968) pp. 115–122, 127–150.
8 The Senster’s code manuscript (8 November 1970): www.senster
.com/ihnatowicz/senster/senstercomputer/index.htm (accessed
17 October 2018).
9 E. Ihnatowicz, Cybernetic Art: A Personal Statement, self-published
(London: 1986).
10 J. Gardner, “Edward Ihnatowicz,” in James Gardner Archives (Uni-
versity of Brighton), unpublished manuscript (8 November 1988).
11 Cybernetic Art of Edward Ihnatowicz, Ihnatowicz family archives,
unpublished film, c. 1980.
Manuscript received 20 December 2018.
AnnA OlszewskA is a visual studies researcher working on
projects connected with the performative function of images
and the history of science. She is the initiator and curator of
the Re:Senster project.
MArek DługOsz is an automation and control system re-
searcher at the AGH in Kraków. He specializes in the design
of autonomous vehicles and is the Re:Senster project control
system designer.
Downloaded from http://direct.mit.edu/leon/article-pdf/54/3/299/1925060/leon_a_01828.pdf by guest on 08 September 2023
Olszewska and Długosz, Senster 305
