Upload
hanhan
View
222
Download
0
Embed Size (px)
Citation preview
Driver-Vehicle-Environmentmonitoring modules
Katia Pagle, ICCS
www.aide-eu.org
AIDE Driver – Vehicle – Environment module: the problem that addresses
The objective is to develop techniques for real-time monitoring of the driver-vehicle-environment state, in order to enable real-time adaptation of the driver – vehicle interface.
www.aide-eu.org
DVE detected conditions
A stored user profile combined with a user’s driving record in terms of user’s preferences and driving behaviour.
Driver profile
The total level of risk concerning the environmental and the traffic conditions (environment type, traffic density, environmental conditions etc.).
Traffic and environmental risk
The driver’s intention, e.g. for a lane change.Driver intent
The physical ability of the driver to drive (fatigue, sleepiness, etc.).
Driver impairment
Cognitive load or shift of visual attention away from the road ahead, induced by an external event or a secondary task.
Driver Distraction
Driver’s “level of availability” to receive and process information, according to the requirements of the current driving task (not available high driving demand).
Driving demand
Explanation DVE parameter
www.aide-eu.org
DVE component
PM
Profile data
Sensor array
DVEMDAE moduleDriver’s availability assessment with respect to the
primary driving taskCAA module
Driver’s Cockpit activity assessment
+detection of driver
intention
DSD moduleDriver’s
physiological and behavioural diagnostic
TERA moduletraffic and
environment risk assessment functions
+detection of driver
intention
DC moduleDriver’s
characteristics and preferences parameters
Functional architecture of DVE modules
www.aide-eu.org
Sensor definition and adaptation
Sensor array
Radar Laser Scannerobject listBUILT-IN
InertialVehicle stateBUILT-INCAN bus
Frontal CameraLaneBUILT-IN
GPS, gyro and digital mapsSynergies with MAPS&ADAS PReVENT
Camera based Eye tracking
AIDE Maps interface
Eye gaze and head positionstereo vision
www.aide-eu.org
Eyelid sensor
The Eyelid sensor is a vision based system which processes the video flow of the driver's face to provide the eye opening and blink behaviour signals.Physically the sensor is composed of a camera including a set of pulsed near infrared (NIR) LEDs placed in the instrument panel linked to a laptop which processes the image flow provided by the camera.
www.aide-eu.org
Map data extraction
MAPS Attributes
Segment Crossing Traffic Sign Speed limits Curvature
Data base
ADASRP S/W
AIDE interface
DVE Module
DVE Module
…
www.aide-eu.org
Traffic Environment Risk Assessment (TERA) module
TERA role:To calculate in real time a total level of risk related to traffic and environmental parameters.To calculate environmental and traffic parameters according to the requirements of the other DVE modules.To estimate the drivers’ intention (e.g. for maneuver of a possible lane change).
Environmentlightweather
Infrastructuretypedistance
For example, existing sensors used by collision-warning systems and Global Positioning System (GPS) measurements of localization combined with a table of corresponding roadway characteristics could be used to help understand the environment outside the vehicle and adapt the HMI accordingly.
www.aide-eu.org
Frontal Radar
Inertial vehicle sensors
Frontal Camera
Map and positioning
data
interfaces
Scenariorepresentation
(data collection)
RiskAssessment(Decision)
CAA driver intention
TERA
TERA algorithms
TERA outputto
ICA and applications
TERA outputto
other DVEmodules
Risk functions
Driver intention (Internal
parameters)
www.aide-eu.org
Driver State Degradation (DSD) module
The DSD is based on the fusion of information provided by physiological and mechanical sensors.
Several diagnostics are calculated: physiological diagnostic, and behavioural diagnostics; then fused with additional in-vehicle information.
Goal: Assess the evolution of the driver state degradation(physical ability).
Behavioral Sensors Lateral position
Learning physiologicaldiagnostic
Physiological sensorsEyelid movement
Behavioraldiagnostic
Diagnostic fusion
Driver state degradation level
Filtering and Featureextraction
In -Vehicle information
Filtering detection and classification
Driver vigilance status level
Abnormal Driving index
BehavioraldiagnosticPhysiological
diagnostic
www.aide-eu.org
Driver Availability Estimation (DAE) module
Goal: Assess the availability of the driver to receive and process an information at a given time
Three levels of Availability / Unavailability:Available (Green): “I can receive and process information”Unavailable (Red): “I drive my car! Please don’t disturb me”Intermediary level (Amber): “You can send information to me, if needed. ButI’m not sure that I will be able to detect or to process it correctly at this time”
www.aide-eu.org
Goal: monitor the activities of the driver and assess visual distraction, cognitive distraction, and signs of lateral manoeuvring intent.
Cockpit Activity Assessment (CAA) module
Lateral manoeuvringintent
Head/eye sensor
Lane position sensor
Standard vehicle sensors
Gaze-world mapping
Vehicle data post
processing
Head/eye data post
processing
Cognitive distraction indicator calculation
Lateral manoeuvring intent detection
Eyes-off-road detection
Visual time sharing detection
Cognitive distraction detection
Sensor data post processing Feature extractionInput Activity assessment Output
Eyes-off-road
Visual time sharing
Cognitive distraction
Lateral manoeuvringintent
Head/eye sensor
Lane position sensor
Standard vehicle sensors
Gaze-world mapping
Vehicle data post
processing
Head/eye data post
processing
Cognitive distraction indicator calculation
Lateral manoeuvring intent detection
Eyes-off-road detection
Visual time sharing detection
Cognitive distraction detection
Sensor data post processing Feature extractionInput Activity assessment Output
Eyes-off-road
Visual time sharing
Cognitive distraction
www.aide-eu.org
Driver Characteristics (DC) module
User profile configuration
agentPreferred interface elements
Disability
Driving experience
Language
Age
STATIC
PA
RA
METER
S
Self-input of profile by the user
INTERFACE
Preferred interface elements
Disability
Driving experience
Language
Age
STATIC
PA
RA
METER
S
Self-input of profile by the user
INTERFACE
Customisationagent
Reaction Time
TLC
TTCDY
NA
MIC
PA
RA
METER
S Reaction Time
TLC
TTCDY
NA
MIC
PA
RA
METER
S
Interface agent
INPUT DC MODULE OUTPUT
• Time of warning (early, normal, late; continuous within ranges)
• Preferred or not-to-be used modality of warning/info
• Intensity (soft, normal, hard; discrete values)
• Adaptation of specific ADAS (applications)
www.aide-eu.org
Thank you for the attention!
Katia Pagle
email: [email protected]: 0030 210 772 3865
fax: 0030 210 772 3557
AIDE Integrated Project web site: www.aide-eu.org