Robot Canvases Developed Utilizing Recreated Robots.

Uploaded on:
Robot Depictions Developed Utilizing Mimicked Robots EvoMUSART '06 Gary R. Greenfield College of Richmond, USA Layout Inspiration Foundation S-Robots Transformative Structure Appraisal Parameters Developed S-Robot Works of art On Self-sufficient Assessment Conclusions Inspiration
Slide 1

Robot Paintings Evolved Using Simulated Robots EvoMUSART ‘06 Gary R. Greenfield University of Richmond, USA

Slide 2

Outline Motivation Background S-Robots Evolutionary Framework Assessment Parameters Evolved S-Robot Paintings On Autonomous Evaluation Conclusions

Slide 3

Motivation “Artistic ability is a long way from an enchantment indefinable embodiment, controlled by a couple and cursed by deconstruction. Maybe it can be considered as a versatile framework , comprising of a specific upgrading plan and low level neighborhood principles or procedures, which have been touched base at through a developmental process.” - Katie Bentley, GA’02 Generative Art Conference, Exploring tasteful example arrangement, pp. 201-213.

Slide 4

Leonel Moura: ArtSBot

Slide 6

J. McCormack: EvoMUSART ‘05 Open Problem #3: “To make EMA frameworks that deliver workmanship perceived by people for its masterful commitment (instead of any simply specialized obsession or fascination).” Open Problem #5: “To make manufactured biological systems where operators make and perceive their own creativity.” My Observation: To perceive their innovative endeavors operators must have the capacity to assess their imaginative endeavors.

Slide 7

Background - V. Ramos and F. Almeida (2000), Artificial insect provinces in advanced picture natural surroundings – a mass conduct impact study on example acknowledgment . - L. Moura and H. Pereira (2002), Artistic Swarm Robots (ArtSBot). - N. Monmarche et al (2003), Interactive development of insect province canvases. - G. Greenfield (2005), Evolutionary systems for insect state compositions.

Slide 8

Monmarche Greenfield

Slide 9

S-Robot Design Loosely displayed after Khepera robots (Binary esteemed) closeness sensor (Tristimulus) shading sensor

Slide 10

S-Robot Specifications Center position (rx, ry). Unit vector course heading (dx, dy). Three forward closeness sensors, and one back nearness sensor, touchy to a spiral separation of 20 units. Forward “field of vision” - 90 deg. to +90 deg. Back “field of vision” - 60 deg. to +60 deg. All closeness sensors are parallel esteemed. Focus mounted tristimulus shading sensor. Two focus mounted “pens” which, when working in coupled, make an imprint five units wide.

Slide 11

S-Robot Commands MOV <arg> - move <arg> units SWI <arg> - swivel <arg> degrees SPD <arg> - set rate <arg> miniaturized scale units per time step SNP <arg> - sense vicinity by overhauling estimations of the nearness vector SNC <arg> - sense updating so as to shade estimations of the shading vector PUP <arg> - pen #<arg> up PDN <arg> - pen #<arg> down Notes: Discrete occasion reproduction decides number of time steps required when attempting to finish a move or when attempting to finish a swivel.

Slide 12

S-Robot “On-Board” Controller Queues a summon arrangement then dozes until succession is executed. Groupings can incorporate “motifs”. on the off chance that (detected red segment == target esteem) qzigzag(q);/* plan “zigzag” theme */q.put(SWI), q.put(- 55); else q.put(SWI), q.put(20); q.put(PDN), q.put(P1); q.put(SPD), q.put(750); q.put(MOV), q.put(12); q.put(SWI), q.put(- 10); q.put(PUP), q.put(P1) q.put(SNC), q.put(0);

Slide 13

S-Robot Testing…Wall Avoidance

Slide 14

…Collisions ...Periodicity

Slide 15

…Pens …Colors …Motifs

Slide 16

Evolutionary Framework GOAL: Using two hand-created controllers, advance beginning positions (sx, sy), where 0 < sx, sy < L, and introductory genuine compass headings d, where - 180 < d < 180, for either TWO or FOUR S-Robots. Lattice Side Length: L = 200. Number of time steps: T = 150,000. Populace Size: P = 16. Number of Generations: G = 30. Substitution: P/2 people utilizing P/4 “breeding pairs”. Recombination: One-point hybrid. Change: Non-elitest (!) point transformation. Winnowing Interval: Every I = 5 eras.

Slide 17

Assessment Parameters Np – the quantity of squares of the framework that were painted. Nb – the quantity of times a S-Robot responded in light of a forward nearness bit set, yet raise vicinity bit clear. Ns – the quantity of a S-Robot responded because of a forward closeness bit set and back nearness bit set. Nc – the quantity of times a S-Robot discovered a wanted shading through shading detecting.

Slide 18

Evolved S-Robot Paintings Fitness F = Np + 1000Nb + 100Ns utilizing two Type A controllers which don\'t make utilization of the SNC shading sense summon.

Slide 19

Fitness F = Np, again utilizing two Type A controllers which don\'t make utilization of the SNC shading sense summon. Eras 10, 20, and 30:

Slide 20

Generations 10, 20, and 30:

Slide 21

Fitness F = Np - 100Ns +1000Nc, utilizing two Type A controllers which don\'t make utilization of the SNC shading sense order. Eras 5, 10, and 15:

Slide 22

Using Type An and B Controllers Fitness F = Np-Nb+100Ns+1000Nc. Eras 0, 15, and 20.

Slide 23

Generations 10, 20, and 30 in the wake of changing one of the chose themes.

Slide 24

Fitness F = Nb*Nc, which chooses for firmly coupled after conduct. Eras 0, 10, and 20.

Slide 25

Fitness F = Nb*Nc + Np*Ns, which again chooses for taking after conduct additionally tries to expand canvas scope. Eras 5, 15, and 30.

Slide 26

On Autonomous Evaluation S-Robots keep up a past filled with their past beginning positions and headings. S-Robots keep up their current Nb, Nc, and Ns values. After T time steps, a chose S-Robot “sweeps” the canvas to calcuate Np. By sharing information, every S-Robot computes the wellness F. By contrasting and past spared wellness values S-Robots chooses which spared genomes to recombine for the cutting edge. By means of a PRNG new genomes are self-produced and S-Robots re-position and re-situate themselves for the following generation’s painting.

Slide 27

Conclusions Fitness capacities can prompt style for advanced S-Robot canvases. S-Robots can aggregately assess their own innovative endeavors. S-Robot practices that (in a roundabout way) impact feel can be advanced.

Slide 28


View more...