World Electric Vehicle Journal,
Год журнала:
2024,
Номер
16(1), С. 9 - 9
Опубликована: Дек. 27, 2024
This
study
presents
an
adaptation
of
the
YOLOv4
deep
learning
algorithm
for
3D
object
detection,
addressing
a
critical
challenge
in
autonomous
vehicle
(AV)
systems:
accurate
real-time
perception
surrounding
environment
three
dimensions.
Traditional
2D
detection
methods,
while
efficient,
fall
short
providing
depth
and
spatial
information
necessary
safe
navigation.
research
modifies
architecture
to
predict
bounding
boxes,
depth,
orientation.
Key
contributions
include
introducing
multi-task
loss
function
that
optimizes
predictions
integrating
sensor
fusion
techniques
combine
RGB
camera
data
with
LIDAR
point
clouds
improved
estimation.
The
adapted
model,
tested
on
real-world
datasets,
demonstrates
significant
increase
accuracy,
achieving
mean
average
precision
(mAP)
85%,
intersection
over
union
(IoU)
78%,
near
performance
at
93–97%
detecting
vehicles
75–91%
people.
approach
balances
high
accuracy
processing,
making
it
highly
suitable
AV
applications.
advances
field
by
showing
how
efficient
detector
can
be
extended
meet
complex
demands
driving
scenarios
without
sacrificing
computational
efficiency.
Sensors,
Год журнала:
2025,
Номер
25(4), С. 1248 - 1248
Опубликована: Фев. 18, 2025
Navigation
systems
are
developing
rapidly;
nevertheless,
tasks
becoming
more
complex,
significantly
increasing
the
number
of
challenges
for
robotic
systems.
can
be
separated
into
global
and
local
navigation.
While
navigation
works
according
to
predefined
data
about
environment,
uses
sensory
dynamically
react
adjust
trajectory.
Tasks
complex
with
addition
dynamic
obstacles,
multiple
robots,
or,
in
some
cases,
inspection
places
that
not
physically
reachable
by
humans.
Cognitive
require
only
detecting
an
object
but
also
evaluating
it
without
direct
recognition.
For
this
purpose,
sensor
fusion
methods
employed.
However,
sensors
different
physical
nature
sometimes
cannot
directly
extract
required
information.
As
a
result,
AI
increasingly
popular
acquired
information
controlling
generating
robot
trajectories.
In
work,
review
mobile
localization
is
presented
comparing
them
listing
advantages
disadvantages
their
combinations.
Also,
integration
path-planning
looked
into.
Moreover,
analyzed
evaluated.
Furthermore,
concept
channel
navigation,
designed
based
on
research
literature,
presented.
Lastly,
discussion
conclusions
drawn.
Sustainable Energy Research,
Год журнала:
2024,
Номер
11(1)
Опубликована: Июнь 27, 2024
Abstract
Many
developing
countries,
particularly
in
Africa
and
Asia,
still
widely
use
traditional
cooking
methods
that
rely
on
solid
fuels
such
as
wood
charcoal.
These
inefficient
polluting
practices
have
severe
health
impacts
due
to
household
air
pollution,
they
contribute
environmental
degradation
through
deforestation
black
carbon
emissions.
This
has
driven
growing
interest
cleaner
more
sustainable
alternatives
electric
(e-cooking),
improved
biomass
cookstoves,
biogas
systems,
modern
fuel
stoves
can
reduce
emissions
consumption
while
providing
a
safer
experience.
E-cooking
emerged
promising
option
sustainability,
benefits,
energy
efficiency,
convenience,
safety,
potential
for
grid
integration,
making
it
alternative
methods.
study
followed
the
PRISMA
guidelines
systematic
reviews
assess
existing
literature
e-cooking
from
1993
2023.
In
addition,
biblioshiny
package
R
software
was
used
perform
bibliometric
analysis
identify
key
trends
evolutions.
The
results
indicate
United
Kingdom,
States,
Japan,
Australia,
China
are
top
five
countries
leading
research.
identified
areas
future
research,
optimising
solar
e-cookers
using
artificial
intelligence
techniques,
integrating
internet
of
things
automation
technologies
e-cookers,
appliances
into
smart
examining
effective
behavioural
change
interventions,
exploring
innovative
business
models.
findings
highlight
need
interdisciplinary
collaboration
among
researchers,
engineers,
social
scientists,
policymakers
address
technical,
economic,
socio-cultural,
factors
influencing
transition
e-cooking.
Sensors,
Год журнала:
2025,
Номер
25(3), С. 856 - 856
Опубликована: Янв. 31, 2025
Autonomous
vehicles
(AVs)
rely
heavily
on
multi-sensor
fusion
to
perceive
their
environment
and
make
critical,
real-time
decisions
by
integrating
data
from
various
sensors
such
as
radar,
cameras,
Lidar,
GPS.
However,
the
complexity
of
these
systems
often
leads
a
lack
transparency,
posing
challenges
in
terms
safety,
accountability,
public
trust.
This
review
investigates
intersection
explainable
artificial
intelligence
(XAI),
aiming
address
implementing
accurate
interpretable
AV
systems.
We
systematically
cutting-edge
techniques,
along
with
explainability
approaches,
context
While
technologies
have
achieved
significant
advancement
improving
perception,
transparency
autonomous
decision-making
remains
primary
challenge.
Our
findings
underscore
necessity
balanced
approach
XAI
driving
applications,
acknowledging
trade-offs
between
performance
explainability.
The
key
identified
span
range
technical,
social,
ethical,
regulatory
aspects.
conclude
underscoring
importance
developing
techniques
that
ensure
explainability,
specifically
high-stakes
stakeholders
without
compromising
safety
accuracy,
well
outlining
future
research
directions
aimed
at
bridging
gap
high-performance
trustworthy
SAE technical papers on CD-ROM/SAE technical paper series,
Год журнала:
2025,
Номер
1
Опубликована: Апрель 1, 2025
<div
class="section
abstract"><div
class="htmlview
paragraph">Autonomous
ground
navigation
has
advanced
significantly
in
urban
and
structured
environments,
supported
by
the
availability
of
comprehensive
datasets.
However,
navigating
complex
off-road
terrains
remains
challenging
due
to
limited
datasets,
diverse
terrain
types,
adverse
environmental
conditions,
sensor
limitations
affecting
vehicle
perception.
This
study
presents
a
review
integrating
their
applications
with
technologies
traversability
analysis
methods.
It
identifies
critical
gaps,
including
class
imbalances,
performance
under
existing
estimation
approaches.
Key
contributions
include
novel
classification
datasets
based
on
annotation
methods,
providing
insights
into
scalability
applicability
across
terrains.
The
also
evaluates
conditions
proposes
strategies
for
incorporating
event-based
hyperspectral
cameras
enhance
perception
systems.
Additionally,
we
address
lack
unified
evaluation
metrics
introducing
qualifiers
assessing
perception,
planning,
control
Finally,
comparison
geometry-based,
learning-based,
probabilistic
methods
navigability
prediction
highlights
importance
multi-sensor
data
integration
improved
decision-making.
These
actionable
recommendations
aim
guide
development
adaptive
robust
autonomous
systems,
advancing
real-world
environments.</div></div>
Designs,
Год журнала:
2025,
Номер
9(2), С. 47 - 47
Опубликована: Апрель 11, 2025
Light
detection
and
ranging
(LiDAR)
sensors
are
critical
for
autonomous
vehicles
that
require
unparalleled
depth
sensing.
However,
traditional
LiDAR
designs
face
significant
challenges,
including
high
costs
bulky
configurations,
limiting
scalability
mass-market
adoption.
By
uniquely
combining
patent
scientometric
analysis,
this
study
screened
188
recent
patents
from
a
dataset
of
more
than
two
million
patents,
uncovering
strategies
to
enhance
capability
reduce
production
costs.
The
key
findings
highlight
the
growing
emphasis
on
solid-state
architectures,
modular
designs,
integrated
manufacturing
processes
as
pathways
scalable
efficient
solutions.
These
insights
bridge
gap
between
scientific
advancements
practical
implementation,
providing
stakeholders
with
clear
understanding
technological
landscape
emerging
trends.
identifying
future
directions
actionable
opportunities,
work
supports
development
next-generation
systems,
fostering
innovation
enabling
broader
adoption
across
other
sectors.
