PS3060: Perception and Action
Term II,  MONDAY 10 - 12 am    (Room 128 Wolfson)

Lecture 5: Bio-Robotics & Neuro-Engineering

Course co-ordinator: Johannes M. Zanker, j.zanker@rhul.ac.uk, (Room 218)


Topics Lecture 5:

  •   the approach & opportunities of biorobotics
  •   orientation and navigation
  •   flying without pilot
  •   limbs as high-tech tools
  •   redefining the boundaries between man and machine 
  •   a variety of questions, a variety of toys


Bio-robotics and neuro-engineering: what is this about?

the dream of robotics: stupid and dangerous work should be done by intelligent machines instead of unreliable people
neuroscience is at the cutting edge of 21st century engineering: robots need to be designed that mimick living systems ('biomimetics', 'neuromorphics')

principles of biological information processing are very successful

another way to look at this new endeavour:         the marriage of kids' dreams and philosophers' nightmares

(see the books of robotics celebrity Rodney Brooks, and many popular movies)


Insect orientation and navigation

robots can be treated as models to describe behaviour (Webb 2001, cf. Braitenberg’s vehicles in lecture 1: ‘synthetic psychology’)
two simple examples:

  • phonotaxis in crickets (Barbara Webb, Stirling): female crickets are attracted by specific songs of male crickets, find the male in its borrow by tracking the sound source
  • navigation and orientation of ants (Cataglyphis) in Sahara: integration of optic flow, sun & polarisation compass, path integration (dead reckoning), orientation relative to landmarks
 >>>  Sahabot 2 
 (Lambrinos, Wehner et al.)


Panoramic Vision  


the richness of panoramic vision can be exploited by using panoramic cameras modelled after insect visual systems:
lens & parabolic mirror
   (Chahl and Srinivasan 1997) 

the polar coordinate images captured with the panoramic camera are 'dewarped' by a computer program to generate conventional cartesian images
         
             the panoramic camera is used in various scientific contexts:

 

  • used in flying robot: gondola Melissa (AILab Zürich) is a test ‘platform’ to study the integration of motion information (flowfields) for estimating flight speed and travelling distance (no proprioceptive information) 
  • used as instrument to study visual ecology: camera mounted on gantry, controlled by microprocessor and stepping motors >> travelling on exactly defined paths : recording flowfields in natural environments (Zeil & Zanker)


Autonomous vehicles: from flies to helicopters

the ultimate goal is to build fast flying, autonomously stabilising and navigating robots: UAV (unpiloted airborne vehicles)
two typical examples:

  • autonomous helicopters solving tasks (Robotics Research Lab, USC): AVATAR (Autonomous Vehicle Aerial Tracking And Reconnassaince), opening the possibility of formation flying through the coordination of multiple autonomous flying robots


Robofly: aerodynamic problems


some science fiction: fly-sized autonomous robots (Berkeley Robotics Laboratory)

surprising complications : unexpected issues arising from specifications of motor system – insect aerodynamics is merely understood, requires complicated aerodynamics (Dickinson 2001)


Autonomous airborne vehicles: political issues

the real goal behind many of these efforts: robotic bombers, etc.

strategic advantages of unpioloted airborne vehicles over cruise missiles

(in first Iraq war it took hours before launching missiles to upload the on-board computer with flightpaths for a given target,
in the second Iraq war … )

‘unmanned combat air vehicles (UCAV) will keep humans in control but out of danger’ (Ashley, Scient. Am. June 2001)
        (this view very much depends on which side you are, j.m.z.)

most funds originate from US Air Force, DARPA, NASA, Office of Naval Research...


Perhaps more comfortable: the soft grip

surgical robots (offer high precision and potential to minimise damage)
require advanced mechanical control of tools (for instance, to handle soft, compliant tissues)
– what has biomimetics to offer?
 
robots mimicking cockroaches (‘bugbots’) are originally designed to walk in unstructured environments with little sensory feedback by robust design of locomotion system, involving a passive rubber spring, like springy resilin-lined joints of arthropods, for passive stabilisation

the same design principle can be extended to the gripper (attached to front limbs)
–  it can be adjusted in stiffness for fine deployment of grip force !
(Harvard Biorobotics Laboratory)

tele-manipulation - remotely controlled hand-like actuators : an emerging field of technology


artificial hand has 5 fingers with 14 degrees of freedom, was built uniquely to test the force feedback glove in a dextrous telemanipulation

Laboratoire de Robotique de Paris

Jet Propulsion Laboratory, managed by the California Institute of Technology, is NASA's lead center for robotic exploration of the solar system (Mars robots!):
Nondestructive Evaluation and Advanced Actuators (NDEAA) Technologies lab

JPL's NDEAA lab produces robotic hand that was established as a platform for artificial muscles:  the arm was made by Dr. Graham Whiteley of Sheffield Hallam University, U.K., the hand was equipped with an actuator by Dr. Giovanni Pioggia, University of Pisa, Italy


Artificial limbs

the same type of technology offers new opportunities for replacement of amputated limbs


conventional prosthesis design

1849 patented by F. Palmer,
improvement of earlier models (increased mobility through artificial knee joint)

to supply 30,000 veterans from the Civil War  
(… technology for yet more victims of wars …)
anthroform arm project : attempt to copy functional anatomy of arm as close as possible – fibreglass bones and pneumatic actuators (muscles), surgical replacement joints, computational models of neural control circuits (University of Washington Biorobotics Laboratory)

there is an obvious demand for intelligent and biologically inspired engineering in prosthesis !

a potential problem for such high-tech prostheses is the attachment to the patient - control from the brain?
issues of mechanical design and pure joint kinematics >> problem of direct control form nervous system
the goal of this development is ‘neuro-prosthesis’: chronic electrode implants would allow to connect artificial limb to nervous system


W. Craelius, Ruttgers: ‘Dextra

the best system so far is an artificial hand tapping into the residual sensorimotor system to activate replacement body parts :

•      pick up electric signals from more proximal                         muscles/tendons
•      learn to control ‘phantom’ fingers
•      independent control of 5 fingers

this demonstrates the successful connection of artificial limb to nervous system - so far to peripheral nerve/muscle : what about the CNS ?


New perspectives of primate-machine interaction

the key animal experiment (Wessberg et al 2000):

remote control
brain >> robot arm
  • chronic implantation of large electrode array (96 electrodes)
  • long-term recordings > correlational analysis with motor patterns
  • transfer data through web to robotics experts
  • drive robot arm with transmitted signals

  • the neuro-scientist (M. Niclolelis)
  • the (owl) monkey
  • the robotic arm

Human brain-machine interaction ? From science to science fiction !

an rather unusual experiment was done by Kevin Warwick in the Department of Cybernetics at the University of Reading
•     Project Cyborg 1.0
•     Project Cyborg 2.0


‘A sophisticated new microelectronic implant has been developed that allows two-way connection to the nervous system. In one direction, the natural activity of nerves are detected and in the other, nerves can be activated by applied electrical pulses. It is envisaged that such neural connections may, in the future, help people with spinal cord injury or limb amputation.’


Food for thought: merging mind and machine

compare the speed of technological development and the increasing relevance of IT-based technologies with speed of evolution…
 
  • growth rates of WWW
  • acceleration of computing speed, software efficiency (neural nets, genetic algorithms) - Moore's law predicts that around 2030 your desk PC will have the same computational capacity as your brain ...
  • novel interfaces between nervous system and electronic devices

combined with a neuroscientific theory of the mind, implemented in ever faster computers,
when will the engine of evolution diffuse boundaries between man and machine,
and generate secondary intelligence that will exceed the intelligence of its creators? (Kurzweil 1999)

enthusiastic popular reception of defining technology of 21st century

Robo sapiens’ predicts (from the turmoil of an exploding technology)
the fusion, or at least close coexistence, of man and (intelligent) machine
– ‘apocalyptic optimism’ ?


Other highlights

  •   Neural networks : for pattern generation,   adaptive systems, learning
  • Chip design from principles of cortical architecture
    neuroinformatics (flowfield processor)

  • self-assembling structure and behaviour
    e.g. HYDRA

and some of this engeneering is don through simulated evolution ('artificial life'): see Brooks 2000


The most popular toys for boys and girls …


Conclusions: (wo)man’s intelligent creatures and companions

from INI 

in this lecture we were looking at :

  •   basic concepts, variety of methods & applications  of biorobotics & neuroengineering
  •   examples of how, by copying from nature,  engineers can extend/repair biological systems
  •   speculations about the future relation between   man and machine 


further reading:


some study questions

download lecture handout


back to index page
last update 25/02/2004
Johannes M. Zanker