Context:
Cal
tech
 aero
space
 professor
 
Soon
-
Jo
 Chung
 
holds
 the
 
'
Bat
 Bot
'
 
flying
 robot
,
 
which
 mi
mics
 the
 flight
 
patterns
 of
 the

 actual
 animal
.
 The
 cover
 of
 Science
 Robot
ics
 featur
ing
 an
 image
 of
 the
 
'
Bat
 Bot
'
 
flying
 robot
.
 "
Bat
s
 use
 more
 than
 40

 active
 and
 passiv
e
 
joint
s
,
 [
along
 with
]
 the
 
flexible
 membran
es
 of
 
their
 
wings
,"
 
Soon
-
Jo
 Chung
 of
 Cal
tech
 
told
 Popular

 Mechanic
s
.
 "
It
'
s
 im
practical
,
 or
 
impossible
,
 to
 in
corporate
 [
all
 40
]
 of
 
these
 
joint
s
 in
 the
 robot
'
s
 design
."
 Or
 as

 biologi
st
 Dan
 Risk
in
 of
 the
 University
 of
 Toronto
 put
 it
 to
 P
BS
,
 "
bat
s
 are
 ridic
ul
ously
 stupid
 in
 terms
 of
 how
 complex

 the
y
 are
."
 "
They
 have
 
a
 
shoulder
 that
 can
 move
 in
 all
 the
 
ways
 that
 an
 insect
 one
 can
,
 but
 then
 the
y
 have
 an
 el
bow
,
 and

 
a
 w
rist
,
 and
 five
 
fingers
 and
 
a
 
thumb
 that
 
controls
 part
 of
 the
 
leading
 edge
 of
 the
 
wing
 
membrane
."
 Chung
 is
 the
 lead

 author
 on
 
a
 paper
 that
 made
 the
 front
 cover
 of
 the
 
latest
 issue
 of
 the
 journal
 Science
 Robot
ics
 (
Ri
skin
 was
 not
 part
 of

 the
 study
)
 in
 
which
 Chung
 and
 his
 team
 
describe
 
their
 design
 for
 
a
 robot
ic
 bat
 that
 
uses
 on
board
 
electronics
 to
 mimi
c

 the
 swe
rving
 and
 div
ing
 of
 the
 real
 animal
.
 It
'
s
 name
 is
 Bat
 Bot
,
 or
 B
2
 for
 short
,
 and
 it
 get
s
 away
 with
 just
 nine
 

joint
s
.
 "
Arg
u
ably
,
 bat
s
 have
 the
 most
 
sophis
t
icated
 
powered
 flight
 mechanism
 
among
 
animals
,"
 the
 paper
 
states
.
 The

 research
ers
 wrote
 that
 the
 complex
ity
 of
 bat
 
wings
 in
 flight
 
drew
 them
 to
 the
 animal
 as
 
a
 model
 for
 
flying
 
robots
.
 "
Bat

s
 can
 do
 some
 
a
bsolutely
 amazing
 things
:
 Be
sides
 
being
 
able
 to
 per
ch
 up
side
 down
 (
which
 is
 trick
y
 if
 your
 initial

 condition
 is
 right
side
 up
 and
 
flying
),
 the
y
 can
 
actually
 
catch
 insect
s
 in
 
their
 
wings
 and
 
carry
 them
 back
 home
.
 The

Input:
 two
 golf
 
balls
.
 With
 
a
 
silicone
 
membrane
 stretch
ed
 over
 its
 carbon
-
fiber
 
s
keleton
,
 
a
 head
 c
ramme
d
 with
 an
 on
-
board

 computer
 and
 sensor
s
,
 and
 five
 micro
-
sized
 motor
s
 
s
trung
 
along
 its
 back
bone
,
 Bat
 Bot
 is
 cap
able
 of
 autonom
ous
,
 fl
apping

 flight
.
 Design
ed
 by
 trio
 of
 robot
ic
ists
 led
 by
 
Soon
-
Jo
 Chung
 at
 Cal
tech
,
 it
 was
 un
veiled
 today
 in
 the
 journal
 Science

 Robot
ics
.
 What
 
makes
 Bat
 Bot
 so
 
remark
able
 is
 just
 how
 
damn
 hard
 it
 was
 to
 mimi
c
 
a
 bat
'
s
 natural
 flight
.
 If
 
flying
 was

 an
 art
 form
,
 bat
s
 
would
 make
 
fixed
 
wing
-
pilot
s
 
looks
 like
 the
y
'
re
 finger
 
painting
.
 That
'
s
 
because
 with
 
each
 
flap
 of
 

their
 
wings
,
 "
bat
s
 use
 more
 than
 40
 active
 and
 passiv
e
 
joint
s
,
 [
along
side
]
 the
 
flexible
 membran
es
 of
 
their
 
wings
,"
 
says

 Chung
.
 In
 addition
,
 bat
s
 take
 
advantage
 of
 
a
 whole
 suite
 of
 other
 hard
-
to
-
im
itate
 b
iological
 
tricks
,
 such
 as
 
bones
 that

 adapt
ively
 deform
 
each
 
wing
-
beat
.
 "
Arg
u
ably
,
 bat
s
 have
 the
 most
 
sophis
t
icated
 
powered
 flight
 mechanism
 
among
 
animals
,"

 the
 robot
ic
ists
 write
 in
 
their
 paper
.
 Bat
 Bot
 perform
s
 four
 main
 components
 of
 bat
 
wing
 movement
 
-
 the
 
shoulder
,
 el
bow
,

 and
 w
rist
 bend
,
 and
 the
 side
-
to
-
side
 tail
 
s
wish
.
 To
 build
 Bat
 Bot
,
 Chung
'
s
 team
 first
 had
 to
 dispens
e
 with
 the
 fantasy

 that
 the
y
 
could
 just
 mechani
ze
 fl
apping
 bat
 
wings
,
 
joint
 by
 
joint
.
 "
It
'
s
 im
practical
,
 or
 
impossible
,
 to
 in
corporate
 [

all
 40
]
 of
 
these
 
joint
s
 in
 the
 robot
'
s
 design
,"
 
says
 Chung
.
 Even
 with
 today
'
s
 most
 
advanced
 robot
ic
 technology
,
 you
'
d

 just
 end
 up
 with
 
a
 
heavy
,
 c
lunk
y
 robot
 that
 
would
 never
 make
 it
 off
 the
 ground
.
 In
stead
,
 the
 trio
 por
ed
 over
 b
iological

 studies
 of
 bat
 flight
,
 
including
 

Targets:
a Idx: 262, Δloglikelihood: 0.510
 extreme Idx: 26597, Δloglikelihood: 0.179
ly Idx: 484, Δloglikelihood: 0.123
 helpful Idx: 15695, Δloglikelihood: -0.805
 2008 Idx: 1508, Δloglikelihood: 2.012
 study Idx: 10380, Δloglikelihood: -0.612
 on Idx: 351, Δloglikelihood: -0.650
 bat Idx: 3569, Δloglikelihood: -1.101
  Idx: 259, Δloglikelihood: 0.294
joint Idx: 29926, Δloglikelihood: 0.633
s Idx: 263, Δloglikelihood: 0.055
 author Idx: 10945, Δloglikelihood: 2.361
ed Idx: 345, Δloglikelihood: 0.010
 by Idx: 455, Δloglikelihood: 0.004
 the Idx: 287, Δloglikelihood: -0.080
 biologi Idx: 31259, Δloglikelihood: 0.532
st Idx: 705, Δloglikelihood: 0.016
 and Idx: 305, Δloglikelihood: -1.090
  Idx: 259, Δloglikelihood: 0.133
Discovery Idx: 79376, Δloglikelihood: -1.635
 Channel Idx: 30449, Δloglikelihood: 0.888
 host Idx: 11285, Δloglikelihood: 0.055
 Dan Idx: 2856, Δloglikelihood: 3.662
 Risk Idx: 57854, Δloglikelihood: 15.165
in Idx: 348, Δloglikelihood: 1.039
. Idx: 260, Δloglikelihood: 0.122
 Sco Idx: 84123, Δloglikelihood: 0.721
uring Idx: 31125, Δloglikelihood: -0.105

 the Idx: 287, Δloglikelihood: 0.636
 studies Idx: 31488, Δloglikelihood: 0.651
, Idx: 261, Δloglikelihood: -0.179
 the Idx: 287, Δloglikelihood: -0.096
y Idx: 276, Δloglikelihood: -0.792
  Idx: 259, Δloglikelihood: 0.637
s Idx: 263, Δloglikelihood: 0.370
ought Idx: 22279, Δloglikelihood: 0.007
 to Idx: 288, Δloglikelihood: -0.372
 understand Idx: 16571, Δloglikelihood: -0.426
  Idx: 259, Δloglikelihood: -0.048
which Idx: 1542, Δloglikelihood: -0.382
 of Idx: 304, Δloglikelihood: 0.491
  Idx: 259, Δloglikelihood: -0.381
these Idx: 3824, Δloglikelihood: -0.499
 40 Idx: 1251, Δloglikelihood: 0.963
  Idx: 259, Δloglikelihood: 0.089
joint Idx: 29926, Δloglikelihood: 0.063
s Idx: 263, Δloglikelihood: -0.000
 the Idx: 287, Δloglikelihood: 0.124
y Idx: 276, Δloglikelihood: -0.403
  Idx: 259, Δloglikelihood: 0.316
could Idx: 3659, Δloglikelihood: 0.145
 dispens Idx: 105308, Δloglikelihood: 0.065
e Idx: 265, Δloglikelihood: -0.001
 with Idx: 514, Δloglikelihood: 0.007
 and Idx: 305, Δloglikelihood: -0.289
  Idx: 259, Δloglikelihood: -0.651
which Idx: 1542, Δloglikelihood: -0.151
 were Idx: 2109, Δloglikelihood: 0.223
  Idx: 259, Δloglikelihood: 0.321
a Idx: 262, Δloglikelihood: 0.229
bsolutely Idx: 53799, Δloglikelihood: 0.281

 vital Idx: 26922, Δloglikelihood: -0.033
. Idx: 260, Δloglikelihood: 0.632
 " Idx: 313, Δloglikelihood: 0.148