►
From YouTube: ROS 2 Security Working Group (2020-11-24)
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
B
Okay,
all
right,
so
I
just
wanted
to
start
out
with
just
making
a
message.
Some
administrative
stuff
we've
been
putting
the
meeting
minutes
in
the
roswiki,
and
I
want
to
move
that
over
to
discourse.
If
nobody
has
any
concerns,
it's
just
the
wiki
page,
just
getting
kind
of
long
and
covers
them,
and
it
seems
like
this
discourse
is
kind
of
the
place
to
post
it.
B
I
was
planning
on
posting
the
meeting
notes
under
one
topic,
and
then
you
know
just
the
meeting
agenda
or
the
the
heads
up
invites
on
their
separate
topics.
Somebody.
B
Yeah,
probably
with
different
posts
and
then
rotate
it
every
quarter
or
every
year.
A
D
That's
good
when
one
thing
I
do
was
in
the
same
post
with
the
announcement
of
the
meeting
and
the
call
for
agenda.
I
then
update
the
op
post
with
the
meeting
minutes
and
recording
that
way.
It's
easily
discoverable
and
it's
searchable
by
the
title
and
tag
I.
D
You
you
just
go
to
the
working
group
tag
for
your
working
group
and
then
you
can
scroll
up
and
down
through
the
the
the.
D
Well,
then,
then,
cross
search
with
the.
If
you,
if
you're
pasting,
the
meeting
minutes
in
the
in
the
discourse
post
itself,
it's
all
searchable,
either
by
google
or
by
the
discord
search
engine
we've.
I've
actually
been
putting
the
meeting
minutes
in
I
get
repo
which
and
then
link
to
the
pr
which
makes
a
little
bit
harder
to
search.
But
at
least
you
find
the
meeting
minutes
and
one
repo
that
you
can
search
locally.
If
need
be.
E
Yeah,
actually
I
was
about
to
suggest
that
maybe
it's
too
nerdy
of
me,
but
like
so
first,
I
like
the
idea
of
moving
them
out
of
the
roster
key,
mostly
because
the
ross
wiki
is
for
roswen
stuff
and
I'm
sure
it
gets
very
confusing
for
people
and
then
like
yeah,
where
to
place
them.
It's
a
it's
a
good
question.
E
A
B
D
B
Okay
cool,
so
so,
unless
anybody
objects
I'll,
look
at
putting
them
in
github
and
go
from
there,
I
just
think
that
the
current
format
is
not
only
it's
going
away,
but
there
it's
roz
one,
but
it's
also
kind
of
cumbersome.
B
And
I
do
not
plan
on
porting
over
any
of
the
other
stuff.
I
just
don't
think
that's
really
worth
it
I'll
just
leave
that
there
so
so
the
next
item,
I
just
wanted
to
mention
the
volvo
remediation
process
that
pr
is
still
open.
I
know
I
got
some
comments
from
mikhail
on
that
still
I'll
address
those
shortly
as
wait
to
see
if
I
had
any
other
comments,
chris
from
or
weighed
in
on
it
as
well
he's
on
the
distribution
list
for
the
vulnerabilities
mailing
list.
F
C
B
More
comments,
please
just
add
them
into
the
pull
request,
and
you
know
we'll
leave
that
there.
So
the
next
thing
I
had
on
the
agenda
is,
I
wanted.
I
thought
we
had
a
really
good
discussion
last
time
about.
You
know
we
started
talking
about
g-turtle
goals.
Hopefully,
everybody's
had
a
chance
to
think
a
little
bit
more
about
that.
So
I'd
like
to
discuss
that
a
little
bit
more,
let
me
throw
out
first
of
all,
just
the
the
five
items
that
I
have
that
are
on
the
table.
B
These
are
things
that
I've
gathered
together
from
things
that
we've
talked
about.
We
didn't
talk
about
all
of
them,
specifically
as
jeep
turtle
goals,
but
I
think
they're
things
like
that.
We
as
a
working
group
are,
are
interested
in
seeing
them
move
forward
specific
to
deter
and
we
talked
about
them.
B
So
the
five
things
I
had
were
the
file
system
list
security
implementation,
so
you
know
doing
a
security
implementation
on
an
architecture
that
doesn't
have
a
file
system.
B
We
talked
last
week
about
the
reference
implementation,
possibly
with
move
it,
and
I
mentioned
in
matrix
I
reached
out
to
one
of
the
working
group
leaders
for
the
open,
manipulation
working
group
and
he
was
really
interested
in
it
both
for
for
the
working
group
as
well
as
for
picnic
robotics.
So
I
think
that
would
be
a
nice
environment
for
us
to
continue
growing
security
tools.
We've
talked
about
quality
enhancements
and
continuing
to
go
after
those
there's
a
small
issue
about
uglifying
the
permission
files
to
shrink
them
down.
B
C
B
A
Is
it
I'm
curious
and
sid,
and
I
have
discussed
this,
but
I
want
to
get
the
inside
of
people
a
little
bit
more
familiar
with
move
it
if
we
do
go
through
this
process
for
that
project,
is
it
something
that
other
people
can
use
to
build
on,
or
is
it
not
really
a
standalone
thing?
Does
that
make
any
sense?
E
Well,
mostly
mostly,
you
can
in
a
sense
like
most
people
would
use
the
move.
Group
interface
of
movies
in
their
project,
and
I
mean
I
haven't
been
using-
I
haven't
even
tried,
move
it
on
roster,
so
maybe,
like
everything,
I'm
doing
is
completely
outdated
and
not
relevant.
E
But
but
yeah
basically
like
the
thing
like
one
of
the
reasons
we
mentioned,
this
project
was
like
a
way
to
have
like
so
we
saw
we
found
some
pitfalls
with
topper
three
and
were
wondering
if
we
could
like
say,
hey
look.
There
is
an
example:
application
where
you
can
use
security,
but
I
see
it's
still
more
as
a
testing
ground
for
both
like
okay,
how
the
permissions
are
gonna
explode.
E
What
other
issues
are
we
gonna
find
how
insane
is
gonna,
be
the
load
on
cpu
to
try
to
encrypt
that
I
don't
know
and
then
to
like
actually
be
able
to
have
a
complex
system
to
test
things.
E
One
of
the
things
we
mentioned
last
time
is
also
the
ability
to
specify
what
to
sign
and
what
to
encrypt
and
like
how
much
like
impact
that
could
have,
and
so
I
don't
necessarily
see
it
as
something
that
like
hey
here,
we
give
you
a
movie
project
with
security,
and
then
you
just
have
to
put
that
on
your
robot
and
it's
going
to
move
and
pick
stuff,
but
but
more
of
like
an
example.
Application
of,
like
hey
like
like
the
total
tree
demo,
like
look,
you
can
use
security
with
the
robot.
E
That's
moving
around
and
now
we
could
say
hey
now.
You
can
like
do
that
with
the
platform
that's
moving
around
and
a
robot
arm
that
picks
that,
and
so
for
me,
that
was
more
like
the
idea
and
to
tie
into,
like
other
words,
that's
going
on
in
srs2.
That
would
also
be
a
way
to
like
experiment
in
like
a
framework
like
movie.
E
Okay,
if
we
were
to
put
no
dl5
in
zero,
like
could
someone
just
like
use
this
movie
like
use,
move
it
and
this
new
node
files
and
just
build
their
thing
on
top
and
have
some
things
that
they
can
secure.
A
E
D
B
A
B
Mean
a
prereq,
for
this
is
actually
being
able
to
run
this
all
in
a
simulator
and
because
I
don't
expect
we'll
all
have
the
hardware
you
need.
D
Yeah,
the
simulator
is
a
little
different,
especially
for
security,
because,
like
it's,
a
sole
provider
and
consumer
of
all
your
sensors
and
actuators,
which
isn't
quite
as
representative,
if
your
hardware
and
sensor
is
distributed
across
many
devices.
E
B
Because
one
of
the
things
that
they've
mentioned
is
that
they're
trying
the
manipulation
working
group
is
trying
to
do
something
similar
with
with
the
reference
hardware.
B
D
It
would
be
pretty
close.
Like
you,
you
know,
you've
the
complexity
of
trying
to
secure
the
stack
either
navigation,
remove
it,
it's
pretty
pretty
challenging
in
itself.
It's
just,
but
I
guess
a
lot
of
these
robot
arm
platforms
are
so
also
are
like
single
in
point
like
the
robotique
robot
arm.
It's
like
really
a
wrapper.
It's
like
one
ross
driver
around
a
rapper,
so
all
the
interfaces
for
joint
motor
controller
or
the
sensing
of
the
camera
on
the
arm
kind
of
comes
from
one
note
anyway.
A
I
mean
we
have,
we
have
a
couple
of
arms,
so
we
don't
really
have
the.
I
don't
think
the
sensor
is
necessary
to
really
take
advantage
of
movement,
so
I
feel
like
either
either
yeah
we
need.
We
need
them
to
sort
of
spec
out
a
baseline
reference
system
that
we
can
get
and
then
secure
that
or
yeah
just
go
simulation.
E
So
I
guess
like
it
depends.
Maybe
we
can.
We
can
like
inspect
the
project
in
like
multiple
like
I
don't
bite
size
in
a
sense
like
I
don't
know
like
it's
unfair,
how
many
resources
would
have.
We
would
have
until
galactic
to
actually
work
on
that
as
well,
and
so
could
like
a
first
version,
which
is
just
like
you
have
a
robot
model.
E
You
have
an
rv's
interface
and
a
movie
interface
innovates,
and
you
can
plan
trajectory
using
moving
planners
and
then,
like
maybe
just
say,
okay,
we
feed
it
a
map
with
known
obstacles,
and
we
have
this
robot
in
that
environment.
We
don't
need
to
simulate
sensors.
We
don't
need
to
like
build
a
map
in
real
time,
just
give
it
an
environment
and
ask
it
to
make
trajectories
in
there.
A
B
So
it
sounds
like
that's
something
that
we're
interested
in.
You
know
and
I'll
say
from
everything
that
I've
gotten
from
the
community
there's
it
just
rose
to
as
a
whole.
There's
it
feels
like
there's
going
to
be
more
and
more
of
a
push
to
make
it
simple.
So
if
we
can
do
the
same
through
this,
I
think
this.
This
helps
demonstrate
how
to
make
security,
simple,
straightforward
out
of
the
box.
B
You
know
that's
what
we're
striving
for
with
frost
too.
I
think
that's
that
makes
this
an
important
kind
of
step
along
the
way,
so
I
can
follow
up
and
continue
to
keep
you
updated
on
matrix.
But
let
me
ask
you
know
when
we
think
about
goals
and
what
we're
what
we
want
to
do.
That
seems
like
there's
a
decent
bit
of
work
in
doing
that
simulation,
work
or
doing
you
know
creating
a
reference
implementation.
B
I
think
the
the
conversation
we
started
about
enabling
security
on
our
platform
without
a
file
system
is
also
a
piece
of
work.
Do
we
want
to
keep
both
of
those
moving?
Quite
frankly,
that's
why
I
haven't
reached
out
yet
to
continue
moving
that
long,
because
I
don't
want
to
bite
off
more
than
we
can
chew.
So
what
do
you
think
about
weighing
the
two
projects
against
each
other?
Can
we
do
both?
Should
we
pick
one.
B
I
only
got
I
got
one
person
from
april
sema,
that's
interested
in
it,
and
then
we
were
gonna.
Try
and
just
actually
have
a
meeting
and
nail
a
few
things
down.
We
didn't
get
a
lot
of
interest
but
kind
of
a
passing
yeah.
That's
that's
a
good
idea,
that's
kind
of
a
road
you
know
put
it
on
the
roadmap
thing.
A
No,
no,
I
just
mean
when
we
can
make.
We
can
make
some
assumptions
about
what
that
is,
you
know,
or
we
can,
or
we
can
have
a
number
of
properties
associated
with
that,
rather
than
rather
than
just
saying,
yeah
build
build
everything
without
a
file
system.
I
mean
remember:
we
talked
about
this
a
couple
like
a
month
ago,
where
that
was
I
mean.
What,
if,
like
environment
variables,
is
not
necessarily
part
of
the
file
system
right,
but
that's
it's
a
similar
type
of
thing.
That's
not
necessarily
supported
on
embedded
platforms.
So
are
you
gonna?
A
E
Well,
it
seems
that,
like
in,
in
both
cases,
I
mean
the
this
is
a
scope,
that's
much
wider
than
security
right,
and
that
would
need
buy-in
both
by
the
people
that,
like
requested,
the
change
which
is
the
micro
ros
folks,
but
also
some
synchronization
and
buy-in
from
open
robotics,
because
we're
basically
talking
about
changing
like
having
a
way
to
define
targets
and
tool
chains
and
build
the
entire
rows
to
stack
for
specific
2
chain.
And
that
seems
like
a
lot
of
work.
But
it
also
doesn't
really
seem
like
too
full
on
our
plate.
D
B
A
B
B
All
right
so
we'll
just
that'll
just
be
you
know
some
post
discussions
and
all
but
no
actual
work
on
that
one.
So
what
about
the
quality
enhancements?
So
we've
got
a
lot
of
things.
I
added
them
to
the
agenda
the
list
of
the
laundry
list
of
quality
things.
I
think
we
came
to
the
conclusion
that
most
of
this
has
to
do
with
documentation,
improving
a
lot
of
documentation
and
so
on.
How
much
do
we
do?
B
We
want
to
actually
explicitly
work
on
that
or
set
goals
for
that,
or
are
we
still
waiting
on
other
dependencies.
A
I
mean
we're
definitely
waiting
on
other
dependencies,
but
that
doesn't
that
doesn't
mean
we
can't
make
progress
on
some
of
these
other
things.
It
just
means
that
it
won't
we're
not
going
to
be
able
to
claim
the
quality
level
at
the
end
of
it
without
our
dependencies.
You
know
so
it's
a
question
of
priority.
I
guess.
E
Yeah,
I
guess
I'm
still,
I'm
still
a
strong
proponent
from
for
improving
documentation,
regardless
of
what
quality
stamp
we
get
on
the
site,
especially
today
is
a
very
good
day
to
have
this
meeting,
because
I
received
like
two
different
emails
from
two
different
like
companies
today
that
were
asked,
as
throws
to
questions
on
private
emails,
which
is
another
thing
that,
like
I
think
we
have
a
like
kind
of
a
problem
on
is
like
people
like
don't
use
online
resources
or
like
don't
use
raw
sensors
or
whatever,
even
if
we
try
to
tell
them
to,
and
so
most
of
these
questions
were
really
based
on.
E
Like
oh,
I
didn't
see
any
example
on
like
how
to
use
this
roster
for
services
and
actions,
or
how
do
you
do
this
or
how
do
I,
like,
I
don't
know,
prevent
someone
from
like
just
setting
an
environment
variable
and
like
disabling
security.
For
my
specific
I
don't
know
oem
components,
and
so
I
think
there
are
a
lot
of
things
that,
like
could
be
very
easily
solved
by
more
tutorials
and
more
documentation
on
the
astro2
repo
or
on
rough
index.
E
D
That's
that's
a
good
idea,
so
the
the
user
goes
through
the
whole.
You
know
hoops
for
learning
about
the
roster
core
features
from
the
demos
and
then
after
they're
sort
of
familiar
they
can
revisit
and
how
to
apply.
F
D
It's
a
tough
thing
like,
I
think
it's
sort
of
wallowing
in
a
whole
bunch
of
beginners
and
then,
if
you're
experienced
enough,
you
like
don't
go
there
and
visit
and
then
the
answer.
So
I
think
it's
suffering
from
not
as
many
experts
answering
questions.
D
There's
also
like
a
lot
a
lot
of
new
questions
or
like
I've,
seen
this
before
so
I
think
maybe
75
questions
I
do
answer.
I
end
up
just
linking
to
the
previous
answer,
but
maybe
discoverability
is
a
bit
hard,
the
site's
kind
of
old,
I
think,
there's
a
ros
discourse
thread
on
updating
it
to
be
like
stack
overflow.
C
D
It
seemed
like,
or
was
discussing
something
about
it
with
the
stack
exchange.
I
don't
know
more
about
that,
though.
E
Yeah
and
in
general
I
think
raw
sensors
has
like
two
issues.
One
is
discoverability
and
like
most
of
the
time
like
you
search
for
like
the
same
problem
as
someone
faced
three
years
ago,
but
because
they
used
grass
kinetic
in
their
question
and
you're
looking
for
us
noetic,
then
you
don't
find
it
or
whatever.
E
So
there
is
definitely
an
issue
there
with
a
lot
of
duplicates,
but
because
it's
hard
to
find
the
original
question
and
then,
if
you
have
like
very
specific
questions,
because
they
are
like
not
that
many
experienced
people
that,
like
answer
on
a
daily
basis
on
our
senses,
your
questions
may
also
go
there
today,
because
the
one
person
that
answers
eighty
percent
of
ross
answer
questions
doesn't
know
about
it.
E
But
but
yeah
in
general,
I
think
if
we
can
provide
better
documentation
so
that
people,
when
they
do
a
google
search
either
like
they
end
on
the
ponderos
index
or
on
the
github
repo.
And
then
I'm
still
encouraging
people
to
go
on
references
and
encourage
us
who
know
about
that
specific
topic.
To
try
to
answer
as
many
questions
as
possible
in
there,
so
that
people
can
find
the
information.
D
B
So
it
sounds
like
we
really
want
to
do
some
work
on
the
reference
implementation
with
movement.
We
really
want
to
do
some
work
on
improving
quality
through
documentation.
There's
one
other.
I
think
kind
of
big
effort
item
that
was
left
is
the
test
failures
and
miguel.
I
think
you've
been
kind
of
pushing
that
along.
E
Was
that
something
like
I've
been
I've
been
fixing
like
issues
that
are
assigned
to
the
working
group
as
they
come,
but
this
one
I
like
didn't
have
time
to
look
into
it
and
it's
a
shaky
test.
So
it's
a
bit
more
like
harder
to
find
and
it's
only
resource
rtps,
which
makes
it
also
harder
to
figure
out.
But
I
think
in
general,
like
we
don't
have
that
many
integration
tests,
and
so
it
would
be
great
if
the
ones
we
have
do
pass.
E
D
E
I
think
the
tests
right
now
are
just
like
getting
getting
a
list
of
nodes
generating
a
key
store
without
effects
for
all
of
these
nodes,
launching
these
nodes
and
making
sure
that
they
do
connect
and
do
receive
messages
with
a
given
maximum
timeout.
That's
pretty
big
and
for
some
reason
like
usually
the
slower
one
for
discovery
is
connects
and
the
timeout
is
basically
like
defined
according
to
that.
E
E
D
D
And
then
it's
failing,
do
you
think?
That's
primarily
because
of
discovery
overhead
or
just
so
maybe
something
else
is
flaky.
E
I
think
I
have
I
haven't
tried
for
a
while,
but
I
think
last
time
I
like
just
like
looked
at
like
how
often
it
was
fading
and
I
basically
found
like
at
least
one
example
of
it
every
night
I
tried
to
replicate
it
on
my
machine.
It
didn't
happen
very
often,
but
I
just
did
the
colcon
retest
until
fail.
I
don't
know
like
20
and
I
could
make
it
happen.
C
E
Sure
yeah
I'll
just
drop
you
my
email
address.
I
mean
it's
everywhere
online,
but
you
would
find
it
that's
just
included.
C
E
Yeah,
I
I
I'll
have
another
look
but
like
I
don't
think
I
don't
think
it
failed
in
cyclone,
since
we
did
the
same
integration
test
in
cycle,
but
I
didn't
look
very
closely
and
I'm
not
receiving
like
good
phone
image.
So
maybe
it's
been
failing
every
night
and
I
don't
know
about
it.
B
So
we're
about
to
the
end
of
our
end
of
our
time
here
so
I'll,
just
roll
that
up
and
just
say
for
for
g
turtle,
things
that
we
want
to
work
on
are
the
movement,
implementation
and
documentation
improving
documentation
on
sros
2..
I
think
those
two
actually
go
hand
in
hand,
and
then
I
think
we
also
want
to
work
on
getting
the
test
failures
fixed.
It
sounds
like
that's
kind
of
a
second
priority,
but
it's
also
also
important
that'll,
just
kind
of
happen
through
continued
discussions.
B
Just
keep
that
in
front
of
us,
so
no
actual
explicit
work
other
than
conversations
about
implementing
sros
2
on
a
system
without
on
on
the
non-file
system
implementation
until
we
get
some
more
from
embedded
or
the
middleware
working
group.
B
So
I'll
summarize
that
put
that
in
a
pull
request
out
to
github
we'll
start
doing
minutes.
That
way
and
please
make
comments
on
there.
Does
anybody
have
anything
else
that
you'd
like
to
discuss
this
week?.
F
F
G
Yeah,
thank
you
yeah.
So
I
added
three
points
at
the
end
of
the
agenda.
I'm
not
sure
if
you
can
see
them
I'll
be
clearly
apologize
for
missing.
That
sorry,
no
worries
I'll
just
be
very
quick,
so
one
of
them
is
first
of
all,
sorry
for
missing
out
last
week,
our
last
last
meeting,
because
it's
a
bit
late
for
me
sometimes
but
yeah
ross
lounge,
I
was
I
was
playing
around
with
it.
I
got.
G
I
got
to
see
the
the
meeting
notes
from
from
the
previous
meeting
and
I
and
I
and
I
managed
to
make
it
work,
but
I'm
not
sure
if
it's
got
access
control
it
doesn't
seem
to
have
it
is.
It
is.
Is
that
the
the
status.
A
Of
it
you're
talking
about
about
this
pull
request.
A
It
does
not
have
access
control
yeah,
so
it's
just
encryption
right
now.
Access
control
should
be
difficult.
We
just
didn't
want
to.
We
just
didn't
want
to
explode
that
pr.
We
wanted
to
start
with
sort
of
our
our
mvp.
If
you
will
okay,
the
review
there
jacob
is
is,
is
pushing
us
toward
having
sort
of
a
plug-in
architecture
there
for
ross
launch,
which
I
don't
think
exists,
so
there
may
be
a
long
path
to
actually
getting
this
landed.
I
I'm
curious
to
see
what
that
turns
into.
G
So
currently,
there's
no
plan
until
this
goes
through.
I
guess
right
to
to
work
on
the
access
control
side.
G
Okay,
yeah
makes
sense.
Then
another
thing
about
the.
I
listened
to
the
to
the
recording,
and
I
I
heard
that
you
guys
are
looking
for
use
cases
like
the
moved
one,
so
I
would
like
to
we're
gonna
use
as
a
security
on
on
the
rms
project
anyway.
So
I
think
that's
a
good
use
case.
If
you
guys
agree
with
that,
and
in
fact
we
already
have
like
a
demo.
E
I
think
that
could
be
a
cool
project
like
I've
been
wanting
to
play
with
it
for
a
while.
I
every
time
I
tried
like
it
was
not
compiling
and
every
time
I
saw
the
issues
they
were
like.
Oh
yeah
wait
for
the
next
three
days
and
we
have
the
dependencies
sorted
out,
so
I
kind
of
gave
up
on
it,
but
if
it's
now
in
like
a
working
state
that
is
like
easily
compilable
usable
and
testable
in
simulation,
that
could
be
an
interesting
use
case.
G
G
G
So
I
don't
know
I
would
like
to
hear,
because
I
have
we
haven't
discussed
much
yet,
but
I
would
like
to
hear
ideas
or
if
you
guys
want
to
think
about
how
how
to
do
that,
because
so
we
have
we're
using
draws
to
communicate,
robots
and
they're,
usually
all
around
a
big
space
right.
So
we
work
with
hospitals,
airport
and
the
robot
is
in
different
places.
D
Yeah
that
you
you,
you
could
use
the
pki
to
generate
a
a
relocation
list.
I
think
that
might
be
a
good
thing
that
the
raw
cli
could
kind
of
maybe
do
for
you.
You
just
tell
it
hey
this
key,
and
this
key
store
is
now
revoked
and
it
auto
generates
a
revocation
list
that
you
could
then
serve
through
a
different
process,
but
maybe
there's
a
convenient
to
do
that
in
python
as
well,
and
the
rosty
cli
can
just
stand
a
little
server
to
do
that.
For
you.
E
F
A
Honey,
marco,
I'm
curious.
I
I
need
to
look
at
your
demos
a
little
bit,
and
maybe
you
can
just
tell
me
if
they'll
answer
my
question,
but
I
I'd
like
to
learn
more
about
your
architecture.
You
know
where
are
your?
Where
is
your
ca?
You
know
where
would
a
crl
be
hosted
that
that
type
of
thing
in
a
way
that
makes
sense?
A
Because
you
can't
just
you-
can't
just
revoke
a
certificate?
You
also
need
to.
I
mean
that'll
just
make
that
robot
stop
working.
Is
that
really
your
end
goal.
A
D
Particularly
on
the
topology
of
the
inter-robot
communication,
is
this
like
a
bunch
of
heterogeneous
robots
or
are
there
like
certain
robots
from
certain
vendors?
I'm
just
thinking,
maybe
there's
a
maybe
a
reason
to
introduce
a
hierarchy
of
cas
yeah
as
opposed
to
a
certain
robot
getting
compromised
a
certain
orc
sub
org
gets
compromised.
You
could
revoke
their
sub.
A
G
G
B
Yeah,
I
think
this
is
actually
marker.
That's
a
really
interesting
use
case.
I
think
that's
a
use
case.
That
is
not
probably
as
critical
in
the
in
the
move.
It
use
case.
I
think
movie
is
probably
more
interested
in
permissioning,
individual
things
and
the
idea
of
revocation
and
how
you
would
actually
structure
it.
Whether
you'd
use
intermediate
cas
and
revoke
groups
of
things
at
a
time
and
have
to
have
a
ca
structure.
B
I
think
it's
a
really
interesting
use
case,
so
I
think
it's
a
yeah
and
I
think,
fleet
management
is
you
know,
security
is
going
to
be
key
in
that
whole
space.
So.
G
B
So
I
gotta
ask
one
one
question
of
the
group:
do
you
guys
mind
if
we
tend
to
go
long
from
time
to
time
we
schedule
the
meetings
for
for
a
half
an
hour
when
we
doubled
them
up
to
twice
a
month,
so
we
often
seem
to
run
long,
especially
when
we
have
good
discussions.
I
hope
that's
not
kind
of
a
red
flag
for
anybody.
C
E
There's
any
questions
that
may
come
up
marco
mentioned
that,
like
it's
he's,
it's
pretty
late
for
him
sometimes,
and
so,
if
he
wants
to
join
on
a
regular
basis,
maybe
we
we
can
and
we
can
discuss
it
on
matrix
or
anything
but
like
maybe
we
can
consider
like
not
running
over
for
people
for
who
it's
very
late
or
like
try
to
adjust
the
times
as
we
did
in
the
past
when
we
have
people
on
like
many
different
time
zones.
B
Yeah
I'll
take
a
note
to
just
open
up
a
conversation
on
matrix
about
meeting
times,
because
you
know
we
were.
We
were
alternating
and
then
I
think
everybody
ended
up
pretty
close
to
time
zones
and
now
we're
back
right
now
and
again.
So
I
just
I
don't
want
to
cut
anybody
out
simply
because
we're
not
meeting
at
the
right
time
so
so
I'll
move
that
over
to
matrix,
then.
E
C
D
Could
you
just
copy
the
links
from
the
messages
and
drop
them
in
a
meeting,
so
we
don't
lose
it.
B
C
B
Okay,
so
that's
all
I
got
if
nobody
else
has
anything
else.
Thank
you
very
much
and
it's
great
conversation
and
for
those
of
you
in
the
u.s.
F
B
All
right,
so,
thank
you
very
much
and
I
guess
we'll
talk
to
you
next,
our
next
meeting
and
I'll
see
you
on
discourse
as
well,
our
own
matrix
as
well,
and
discourse
and
discourse
yeah.