A Biologically-Inspired Algorithm for Interception I C O N S 2 0 1 - - PowerPoint PPT Presentation

a biologically inspired algorithm for interception
SMART_READER_LITE
LIVE PREVIEW

A Biologically-Inspired Algorithm for Interception I C O N S 2 0 1 - - PowerPoint PPT Presentation

A Biologically-Inspired Algorithm for Interception I C O N S 2 0 1 9 Frances S. Chance Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly


slide-1
SLIDE 1

Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

A Biologically-Inspired Algorithm for Interception

I C O N S 2 0 1 9

Frances S. Chance

SAND2019-7965 C

slide-2
SLIDE 2

Interception

1

Common behavior in animals…

(Image: Westend61/Getty) (Image: U. Oxford) (Image: Wikipedia) from Lin & Leonardo 2017

Still need solutions for man-made platforms…

from www.scienceabc.com

slide-3
SLIDE 3

Dragonflies as inspiration for an interception algorithm

2

(Image: Westend61/Getty)

Ø Dragonflies intercept prey when hunting Ø Good at it (90-95% capture rate)

from Lin & Leonardo 2017

Time scales of dragonfly interception computation is also very fast Ø Latency to react to prey steering: 50 ms

slide-4
SLIDE 4

3

Dragonflies as inspiration for an interception algorithm

Dragonflies maintain prey at a particular eye-position (foveation) during approach Is target-image slippage on eye enough information for a robust interception algorithm?

from Lin & Leonardo 2017

slide-5
SLIDE 5

Simulating a dragonfly

4

Ø dragonfly eye simulated as a 2D screen Ø fovea at the center of the eye/screen Ø dragonfly maneuvers to keep prey-image on fovea Ø dragonfly and prey move at same maximum speed (unrealistic but more challenging) Ø no restrictions on maneuverability target dragonfly fovea / target image

slide-6
SLIDE 6

Simulating a dragonfly

5

dragonfly-centered reference frame physical-space reference frame Dragonfly maneuvers to keep prey-image at eye-center (fovea)

slide-7
SLIDE 7

Simulating a dragonfly – target held at eye center

6

dragonfly-centered reference frame physical-space reference frame Dragonfly maneuvers to keep prey-image at eye-center (fovea)

Classic pursuit behavior (dragonfly heads straight towards prey) Viable interception strategy except without speed overmatch can fall into ”endless pursuit”

*note: videos replaced by screenshot in PDF *note: videos replaced by screenshot in PDF

slide-8
SLIDE 8

Proportional navigation

aka CBDR (constant-bearing decreasing-range) or parallel navigation

7

Generates the geometrically shortest interception trajectory Dragonflies use a form of proportional navigation to intercept their prey dragonfly prey

slide-9
SLIDE 9

Can proportional navigation be implemented using prey-image slippage on the eye?

8

Ø correct “fixation spot” is calculated based on the trajectory of the prey (relative to dragonfly trajectory) Ø dragonfly maneuvers to keep prey-image on fixation spot Ø dragonfly and prey move at same maximum speed (unrealistic but more challenging) target dragonfly calculated fixation spot

slide-10
SLIDE 10

Can proportional navigation be implemented using prey-image slippage on the eye?

9

Dragonfly maneuvers to keep prey-image at fixation spot (calculated) dragonfly-centered reference frame physical-space reference frame

Potentially more robust strategy if prey has similar speed capabilities However, vulnerable to rapid evasive prey maneuvering

*note: videos replaced by screenshot in PDF *note: videos replaced by screenshot in PDF

slide-11
SLIDE 11

What if the fixation spot isn’t perfect?

10

Dragonfly maneuvers to keep prey-image at fixation spot (with noise added)

Currently working on model for online adaptation of fixation spot (will be applicable for incomplete sensor information, evasive prey, error corrections)

*note: videos replaced by screenshot in PDF *note: videos replaced by screenshot in PDF

slide-12
SLIDE 12

Summary

11

Ø Buliding a model of dragonfly-prey interception Ø Focus on how dragonflies use visual input to calculate interception trajectories Ø Dragonfly maneuvers to maintain prey-image on a particular fixation spot

from Lin & Leonardo 2017

Ø Pursuit behavior if fixation spot is in center of eye Ø Proportional navigation if fixation spot is calculated based upon prey trajectory Ø Ideal strategy may be some hybrid between pursuit and proportional navigation (if prey is evading)

slide-13
SLIDE 13

The End

Questions? Email fschanc@sandia.gov