An Autonomous Wheelchair at an Art Gallery
Beurs van Berlage, Amsterdam, 16 December 2009 to 8 March
2010
Our robotic wheelchair has been exhibited at
Niet
Normaal in Amsterdam from December 2009 to March 2010. See a
Youtube video
or a local copy (WMV) (thanks to Aram
Voermans for filming and editing).
Institute of International Visual Arts (INIVA), London, 2008
The autonomous wheelchair "Psalms" commissioned
by the late Donald Rodney in 1997 has been exhibited at the exhibition
"Donald
Rodney In Retrospect". Curated by Sebastian Lopez and Keith Piper. Produced
by Iniva (Institute of International Visual Arts).
 |
 |
|
Psalms and
Britannia Hospital 3 (Installation view of Donald Rodney in Retrospect) |
Installation
view of Donald Rodney in Retrospect Photograph © Thierry Bal, Courtesy of
Iniva at Rivington Place |
South London Gallery 1997
An autonomous wheelchair has become a piece of art named "Psalms"
as a part of an exhibition entitled "Nine Night in Eldorado" by
Donald Rodney, held at the South London Gallery until
the 12th October 1997 (Fig 1,2,3 and wc1.mov).
"...Our fear of automata is again harnessed in Psalms, as the empty
wheelchair courses through its various trajectories on a sad and lonely
journey of life, a journey to nowhere. Its movements repeat like an ever
recurring memory, a memory of another life and another journey,
that of Donald Rodney's father..." (Exhibition brochure, Jane Bilton.)
|
|
|
Fig.1. A part of the exhibition.
(click on the images to see
larger versions)
|
Fig.2. Waiting for a conversation to end
before resuming its walk...
...or taking part ?
|
Fig.3. A lonely presence after the closure.
|
The wheelchair was modified at the School of Computing of the
University of Plymouth to perform a repeated sequence of circles,
spirals and figures of eight, as specified by the artist Donald Rodney,
himself disabled by sickle cell anaemia.
The wheelchair uses 8 sonar sensors (fig.4) , shaft-encoders, a video
camera
and a rate gyroscope to determine its position. A neural network using
normalised RBF nodes encodes the sequence of 25 semi-circular sequences
of positions forming the trajectory (fig.5).
The control system comprises a laptop PC 586 running a control
program written in CORTEX-PRO, and linked to a Rug Warrior board built
around the 68000 microcontroller.
|
|
|
Fig.4. Orientation of the sonar sensors. The sensors are mainly oriented
in front for safety reasons. Sensors 1,3,5 and 8 are also used for
self-localisation purpose when the wheelchair is oriented parallel to a wall.
|
Fig.5. Neural network based on normalised RBF nodes encoding a sequence
of sequences of positions (a figure of eight in this example). Layer L1
comprises input nodes representing the x,y position of the wheelchair.
Layer L2 comprises usual RBF nodes centred on successive positions. Nodes
in layer L3 perform the normalisation so that their ouputs represent
positions in the same form as nodes in layer L1.
More details on normalised RBF nets can be found in Althofer and Bugmann
(1995), Bugmann (1996), Bugmann et al. (1998).
|
Contributors:
The control system of the wheelchair was built by:
- Dr.
Guido Bugmann (School of Computing, University of Plymouth)
- Kheng Lee Koay (School of Computing, University of Plymouth)
- Dr. Nigel
Barlow (School of Computing, University of Plymouth)
with the help of Steve Hill
(SoC) and Prof. Mike
Denham (SoC),
and the support of Technical Services of the University of
Plymouth.
Financial support was provided by
- A Grant of the Henry Moore Foundation to the South London
Gallery.
- The Mobile Robotics Laboratory in the Centre for Neural
and Adaptive Systems (Prof. Mike
Denham) of the School of Computing of the University of Plymouth (UoP).
We gratefully acknowledge material and technical support by:
- Plymouth Disability Equipment Centre
- British Aerospace Systems & Equipment, Plymouth
- Penny and Giles Drives Technology Ltd, Christchurch BH23 4HD,
UK.
- Paul Robinson (School of Electronic, Communic. & Electrical
Eng., UoP)
We gratefully acknowledge the following people for their very helpful
advice:
- Dr. Alan Simpson (School of Electronic, Communic. &
Electrical Eng., UoP)
- Peter Nurse (School of Manufact., Materials & Mech.
Eng., UoP)
- Peter Frere (Lucas Advanced Engineering Centre,
Birmingham)
- Dr. David
Keating (Dept. Cybernetic, University of Reading)
We are very grateful to Mike Phillips
(School of Computing, University of Plymouth) for
initiating this project by establishing the initial link between the artist and the builders.
Many technical solutions in the control system of the wheelchair are
based on sub-systems
developped during the projects realised by following students:
Vincent Onillon, Priska
Shönborn, Antonio Rodriguez Perez, Jose Dorado, Petr Bunus, Derrick Tapscott.
References:
"Planning
and Learning Goal-Directed Sequences of Robot-Arm
movements" (41411)
Althöfer K. and Bugmann G. (1995)
Proc. of ICANN'95, Paris, Vol. 1, 449-454.
"Note
on the Use of Weight-Averaging Output Nodes in RBF-Based Mapping nets"
(60120)
Bugmann, G. (1996)
Research Report CNAS-96-02
"Stable Encoding of Robot Trajectories using Normalised Radial Basis Functions: Application to an Autonomous Wheelchair" (83,209),
Bugmann G., Koay K.L., Barlow N., Phillips M. and Rodney D. (1998)
Proc. 29th Intl. Symp. Robotics, 27-30 April, Birmingham, UK.
"Normalized Radial Basis Function Networks" (Preprint, 179,885)
Bugmann, G. (1998)
Neurocomputing (Special Issue on Radial Basis Function Networks), 20, pp. 97-110.
(ISSN : 0925-2312)
"Stable
Encoding of Robot Trajectories using Normalised Radial Basis Networks:
Application to an Autonomous Wheelchair." (PDF 337KB)
Bugmann G., Koay K.L., Robinson P. (2006)
The International Journal of Vehicle Autonomous Systems (IJVAS),
Special issue on "Computational Intelligence and Its Applications to Mobile
Robots and Autonomous Systems". Vol. 4 No 2-4. PP. 239-249.
Guido Bugmann, Jan 2010