13. Software Reference Architecture
Visual
Odometry
ADAPTIVE HMI
ENGINE
Visual
Search
Weather
PROACTIVE
Distance to
destination
DECISION
ENGINE
To achieve context definitition and proactivity we use a 4 layer
Cruise Range
software stack
CONTEXT
ENGINE
Location
sensor
OBD Fuel
Level
OBD RPM
AGGREGATOR
Reference Architecture
OBD Engine
Temperature
OBD Speed
14. Reference Architecture: Aggregator
CONTEXT
ENGINE
Cruise Range
The Aggregator component collects sensor data, aggregates them
according to specific rules, and pushes them to the context engine
AGGREGATOR
Location
sensor
OBD Fuel
Level
OBD Speed
15. Reference Architecture: Context Engine
PROACTIVE
DECISION ENGINE
m_ContextAction0= "Show Default Panel"
The Contextm_ContextAction1= "Showthe ontology rules to the sensor
Engine applies Parking Warning"
m_ContextAction2= "Show Fuel Warning"
values and generates lists of facts
m_ContextAction3= "Show POI Available Warning"
m_ContextAction4= "Show VS Match Warning"
Context Facts
Ontology
Rules
CONTEXT
ENGINE
16. Reference Architecture: Proactive Decision Engine
The Proactive Decision Engine is composed by separated subengines communicating via task-board
Adaptive HMI Engine
Proactive
Decision Engine
Taskboard
Navigation
Manager
POI
Manager
Panel
Manager
Warning
Manager
17. Reference Architecture: Proactive Decision Engine
Each sub-engine filters its specific category of facts and creates and
ordered (priority based) list of facts (actions) to be shown in the HMI
Adaptive HMI Engine
PDE takes “decisions”
between possible
solutions
Show
default panel
Show fuel
warning
Proactive
Decision Engine
Navigation
Manager
POI
Manager
Panel
Manager
Warning
Manager
18. Reference Architecture: Adaptive HMI Engine
Finally, the Adaptive HMi Engine selects the proper modality
Show fuel
warning
Show default
panel
AHE selects
modality
Adaptive HMI Engine
Show
default panel
PDE takes
“decisions”
Show fuel
warning
Proactive
Decision Engine
Navigation
Manager
POI
Manager
Panel
Manager
Warning
Manager
19. Database
Remote database manages:
•
•
•
Normal Points of Interest
Visual Search
Multimedia information ofr Augmented Reality
The database is locally buffered when
a specific route is selected, to avoid
connection issues during the trip
24. Results
Visual Odometry algorithms are used when the GPS signal gets lost in
urban canyons, keeping the car position on the map up to date
25. Results
Visual Search algorithms find a visual match on what the camera is
shooting, allowing specific POI information to be delivered when
actually facing a meaningful building