Embodied Artificial Evolution: the Next BIG Thing? by A.E. Eiben

604 views

Published on

A.E.Eiben's presentation slides for his Awareness Virtual Lecture December 2011

Published in: Education, Technology, Spiritual
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
604
On SlideShare
0
From Embeds
0
Number of Embeds
30
Actions
Shares
0
Downloads
5
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Embodied Artificial Evolution: the Next BIG Thing? by A.E. Eiben

  1. 1. Embodied Artificial Evolution: the Next BIG Thing? Gusz Eiben VU University Amsterdam <ul><li>Popular version in the </li></ul><ul><li>TED talk: &quot;Tech Kangaroos: Evolution at Work ” </li></ul><ul><li>Search for “Eiben” on youtube </li></ul>
  2. 2. Past, Present, Future <ul><li>19th century: </li></ul><ul><li>evolution is a theory , an explanatory concept that helps us understand things </li></ul><ul><li>20 th century: </li></ul><ul><li>evolution is a tool , an active, creative concept that helps us produce things (solutions for problems) in digital spaces </li></ul><ul><li>21 st century: </li></ul><ul><li>evolution is a tool , an active, creative concept that helps us produce things (solutions for problems) in physical spaces </li></ul>
  3. 3. WETWARE HARDWARE EVOLUTION EVOLUTION Embodied Artificial Evolution SOFTWARE Biosphere Evolutionary Computing
  4. 4. Evolutionary Computing lessons <ul><li>+ </li></ul><ul><li>Evolution can solve problems we don ’t fully understand and cannot clearly specify </li></ul><ul><li>Evolution can cope with changing situations </li></ul><ul><li>Evolution can come up with original, unexpected solutions (that can be reverse-engineered) </li></ul><ul><li>– </li></ul><ul><li>Designing good EAs can be difficult (representation, operators, parameters) </li></ul><ul><li>No good theory </li></ul><ul><li>Could take too long (can change with new hardware) </li></ul>
  5. 5. Evolutionary computing and embodiment <ul><li>Evolution in digital space, result in digital space </li></ul><ul><li> e.g., time tables, consumer models, robot controllers </li></ul><ul><li>EVOLUTIONARY OPTIMIZATION </li></ul><ul><li>Evolution in digital space, result in physical space </li></ul><ul><li> e.g., jet nozzle, satellite boom, P Bentley ’s coffee table </li></ul><ul><li> EVOLUTIONARY DESIGN </li></ul><ul><li>Evolution in physical space, result in physical space </li></ul><ul><li> EMBODIED ARTIFICIAL EVOLUTION </li></ul>
  6. 6. Embodied Artificial Evolution <ul><li>Old: Evolution of code (blueprint) </li></ul><ul><li>New: Evolution things </li></ul><ul><li>Takes place in physical objects </li></ul><ul><li>Reproduction = real birth = new object made, survivor selection = real death = object deleted </li></ul><ul><li>Fitness is environmental or hybrid environmental and task-based </li></ul><ul><li>Reproduction and survivor selection are decoupled, autonomous decisions, distributed management </li></ul>
  7. 7. Application examples <ul><li>Mars explorers, artificial pets, robot companions, … </li></ul><ul><li>Functional organisms, medical nano-robots, personal virus scanners, … </li></ul><ul><li>Personal replicators (networked, evolutionary, Internet of Things ) </li></ul>
  8. 8. Active vs. passive entities <ul><li>Two kinds of artefatcs </li></ul><ul><ul><li>Active: needs controller to govern its activities (e.g., robot) </li></ul></ul><ul><ul><li>Passive: needs no controller (e.g., smartphone cover) </li></ul></ul><ul><li>In biological terms </li></ul><ul><ul><li>Active: needs body + mind </li></ul></ul><ul><ul><li>Passive: only body </li></ul></ul>
  9. 9. <ul><li>Different approaches, under different umbrellas </li></ul>Physical medium, embodiment ? Bottom-up, chemistry Top-down, biology Mechatronics, plastics
  10. 10. A new game for “passive” stuff Manufacturing Testing (Re)design Embryonic development Evaluation Selection Reproduction propagule blueprint
  11. 11. A new game for “active” stuff <ul><li>Robots </li></ul><ul><li>very small to big </li></ul><ul><li>programmable </li></ul><ul><li>controllable </li></ul><ul><li>sensors/actuators/controllers </li></ul><ul><li>Cells </li></ul><ul><li>very small </li></ul><ul><li>self-replicating </li></ul><ul><li>self-repairing </li></ul><ul><li>self- … </li></ul>WETWARE HARDWARE Sweet spot
  12. 12. Grand challenges <ul><li>Body type </li></ul><ul><li>Reproduction - how to start </li></ul><ul><li>Kill switch - how to stop </li></ul><ul><li>Evolvability and evolution speed </li></ul><ul><li>Process control and methodology </li></ul><ul><li>Co-evolution of body & mind </li></ul><ul><li>Lifetime learning (self-awareness, mind must match body) </li></ul>
  13. 13. Closing remarks <ul><li>Evolution in the real world is different – matter matters </li></ul><ul><li>Computation in this new paradigm is different </li></ul><ul><ul><li>Non-von-Neumannian architecture for information processing/computing </li></ul></ul><ul><ul><li>Redefines ICT, e.g., “ program ” , “ programming ” , software engineering, testing, validation, ect. </li></ul></ul><ul><li>Big questions are implied: </li></ul><ul><ul><li>Scientific (fundamental issues regarding evolution, notion of Life) </li></ul></ul><ul><ul><li>Technological (feasibility) </li></ul></ul><ul><ul><li>Applicability (remember IBM in the 1940ies) </li></ul></ul><ul><ul><li>Desirability (ethics, think of GMOs) </li></ul></ul><ul><li>Different communities, unifying vision / umbrella needed </li></ul><ul><li>Next BIG thing ? </li></ul>

×