Autonomous Navigation and Mapping System for MRPTA Rover

fangscaryΤεχνίτη Νοημοσύνη και Ρομποτική

13 Νοε 2013 (πριν από 3 χρόνια και 10 μήνες)

136 εμφανίσεις

Autonomous Navigation and Mapping System for MRPTA R
over


V.

Polotski, F. Ballotta
, J. James


The Micro
-
Rover Platform with Tooling Arm (MRPTA) project was initiated and financed by the
Canadian Space Agency to develop a small format robotic system
for
use in analog planetary
missions. The MRPTA system
accepts interchangeable instrument payloads
,

provides
both
teleportation and autonomous navigation capabilities
,

and
incorporates

a command language
used to define

mission
s
.
In this paper we address the
onboard
navigation and mapping system

functions used for autonomous operation.

Mechanical design, low
-
level motion control and
teleoperation
are to be addressed elsewhere.

Although
developed
specifically
for

planetary exploration
the navigation system
is
highly

suitable

to
large
number

of

application

domain
s

and could be

of
particular

interest
for mining

and construction

applications
.


The a
utonomous navigation system is capable
of

mov
ing

the
platform to predefined position
s
/orientation
s

while continuously

ma
pping the are
a

it is about
to travel,
assessing the traversability of th
at

terrain and choosing

most appropriate path to
fo
l
lo
w.

The s
ystem design
is based on the
JAUS
(Joint

Architecture for Unmanned Systems)
model

and

consists of interconnected
software
components.
Core

navigation

components are

the
Pose

Estimator (PE
),

the
Terrain Evaluator (TE),
the
Map Manager (M
M
),
and
the
Path

Planner (PP).


Using

a

strictly minimal sensor configuration consisting of an azimuth gyroscope,
an absolute
inclino
meter and
wheel/track odomet
ers
,

t
he PE performs e
xtensiv
e fusi
ng
of
available data
and

provide
s

reliable

pose estimates
even in cases

of pronounced
platform
slippage
.

The
TE
controls

a laser scanner located on
a

tilt axle
to

sweep

the area in front of the platform

and
provide point
-
cloud data to the MM
.
The
MM

employs

a number of

experimentally
-
determined,

tunable
parameters
to
construct
a

“traversability grid”

(TG) of the surrounding
area.

TGs

obtained along the platform motion
path
a
re
continuously integrated
into
topology
and traversability

map
s

maintained by
the

M
M
. Th
ese

map
s

are

used by
the
PP
to com
pute
motion

path
s

for

execut
ion

by
the
motion control

components
.
All autonomous navigation

components
run

concurrently

and

PP
-
c
alculated
motion

path
s

are

considered valid

until
the
platform either deviates from the intended path or
new

map
data invalidates the current plan

--

in these cases motion trajectories are re
-
planned.

The on
-
board a
utonomous navigation
software is

impl
emented in C/C++
running on
L
inux
.
The
system provides soft real
-
time behavior and
has been extensively tested in various locations
including CSA Mars Yard
. The system

was

successfully demonstrated

and
delivered to Canadian
Space Agency in June 2012.