nicole van nes (swov) project coordinator
TRANSCRIPT
Naturalistic Driving in Europe
Nicole van Nes (SWOV) Project Coordinator
UDRIVE versus SHRP2
UDRIVE Large scale European Naturalistic Driving study
Person cars 4 OS, 120 in total France, Germany,
Poland, UK 30 vehicles per OS
Trucks 1 OS, 50 in total
Netherlands
PTW’s 2 OS, 40 in total
15 vehicles in Austria 25 vehicles in Spain
1 O
Project timeline
The Data Acquisition System (DAS)
MobilEye – smart machine vision
Cameras Position – Cars
Forward cameras Feet camera Face camera Driver’s action camera Passenger compartment camera Right blind spot camera*
Cameras Position – Cars
Forward cameras Feet camera Face camera
Driver’s action camera Passenger compartment
camera Right blind spot camera*
orwFo
Cameras Position – Cars
•
•
•
•
•
•
Cameras Position – Trucks
Forward cameras Feet camera Face camera (just 1) Driver’s action camera Passenger compartment camera Blind spot camera*
Cameras Position – Trucks
Forward cameras Feet camera Face camera (just 1) Driver’s action camera Passenger compartment camera Blind spot camera*
Cameras Position – Trucks
•
•
•
•
•
Cameras Position – PTW
Forward cameras Face camera Side cameras Rear camera
Cameras Position – PTW
Forward cameras Face camera Side cameras Rear camera
as
Cameras Position – PTW
•
•
•
•
Study design
• Participant stratification • Vehicle types • Recruitment progress
Participant stratification • Cars: Multi-driver selection to enlarge sample size
• Trucks: availability of driver and vehicles depends on participating fleets
Vehicle types 3 cars types: • Renault Clio 3 (small car) • Renault Clio 4 (small car) • Renault Mégane 3 (Medium-sized family car) 1 PTW type: Piaggio Liberty 2 Truck types: Volvo, medium sized for city deliveries
Progress on participant recruitment
Main research areas
• Crash causation and risk • Everyday driving • Distraction and inattention • Pedestrians and cyclists • Motorcycle behaviour • Eco-driving
Crash causation and risks • Crash causation (“holistic”)
– How can contributing-factor chain schemas be applied to naturalistic road user data?
– What are the factors that contribute to the occurrence of safety critical events for lead- vehicle and intersection conflict scenarios for cars and trucks?
– Are there driver/vehicle/environment factors that frequently occur together in a safety critical events?
• Risk calculation – What are the risks of different driver behaviours? – Is there a difference in the driving related risks when stratified by road
type, vehicle type and country? – What is the risk of disregarding safety precautions?
Everyday driving Descriptive analysis of everyday driving • To what extent are driver factors associated with risky
behavior? • To what extent are environmental factors associated with
risky behavior? • To what extent are driver assistance systems used? • To what extent are seatbelts used? • How does traffic culture influence driving behavior?
Distraction and inattention • Attention selection mechanisms
– Which perceptual cues reliably capture attention and trigger avoidance maneuvers in SCEs?
– Why do the reactive attention capture mechanisms, identified in RQ1, sometimes fail and lead to crashes?
– What factors determine how drivers proactively allocate their attention in anticipation of how a driving situation will unfold and why do these proactive selection mechanisms sometimes fail?
• Involvement in secondary tasks – What are the key factors influencing the willingness of drivers to deliberately
engage in secondary tasks such as phone conversation, dialing or texting? – How do drivers adapt ongoing secondary task activities to the evolving
driving situation? – To what extent can an individual’s willingness to engage in secondary tasks,
and its effects on risk and driving performance, be predicted from psychological tests?
Pedestrians and cyclists • Drivers interacting with cyclists and pedestrians
– What characterizes Safety Critical events (SCE) involving motorized traffic and cyclists/pedestrians at intersections?
– How do car drivers behave at intersections in urban areas where they might encounter cyclists/pedestrians (in normal conditions, not SCE).
– Which ‘external’ factors (e.g., intersection design) modify those behaviours?
• Are the VRU related SCE’s identified by the Mobile Eye system (warnings) correct, relevant, reliable and properly timed?
Motorcycle behaviour
• PTWs behavior (based on riding data) – What characterizes Safety Critical events (SCE) involving PTWs? – Which circumstances (rider, infrastructure, trip) have impact to
SCE occurrence? – What is riders speed choice in relation to the speed limit and the
situation? – What characterizes looking behavior of PTW riders in left turn
maneuvers?
• Behavior towards PTWs (based on driving data) – What is the role of timely perception of a rider by drivers?
(conspicuity)
Eco-driving • Effects of driving styles on eco driving
– Does the vehicle power-to-mass ratio affect the driving style? – How much do drivers deviate from the speed limit in free flow situations,
and why? – Is eco-driving and safe driving correlated, through increased anticipation
of road infrastructure and traffic situations?
• Potential effect of eco-driving – When do drivers brake and is it necessary to brake in each instance? – Is eco-driving a visible characteristic of certain drivers? – Do drivers shift gear to avoid high engine speeds and high fuel
consumption?
Other interesting research topics
Outside the UDRIVE project • Infrastructure • Drowsiness • Traffic management • Use of ADAS systems
Analyses
• Pre-processing and data enrichment
• Preliminary Analyses Plan • Safety Critical Events
Pre-processing and data enrichment • Map matching
– Enrich GPS data with e.g. • Speed limits • Road type • Intersection type
• Video annotation – Central annotations
• By students • Driver ID • Classification of SCE candidates into safety relevant categories • Base annotations of SCEs • Location based annotation of required variables (eg intersection type) • development of codebook
– Local annotations • Analyses partners • Detailed coding, e.g. eye movements, details about the situation.
Preliminary Analyses Plan Main issues: • Strategies for triggers • Development of algorithms for data selection • Definition of Safety Critical Event
– International discussion – New definitions to be developed for VRUs
• Strategies for annotation • Development of annotation code book
– Use SHRP2 annotation schemas as a basis, but adapt some and add several other variables
Data Sharing and Data Protection Concept
• Data available for post project research
• Data access for non-partners
Remote access to UDRIVE data • Analysis performed remotely on the
common dataset
• Central Data Center (CDC) • located at SAFER • Hosts all data
• Local Data Centers (LDC) • located at DLR, Volvo and CEESAR • Hosts part of data
• Partner Data Center (PDC) • Partner can download data and
become PDC
Post-project access to UDRIVE data Infrastructure for remote access and data protection • Third parties can remotely access all
data from CDC, except Personal Identity Data (PID)
• PID (video, GPS) is only accessible at the premises of CDC or PDC
• Third parties must apply to analyze data and adhere to the Data Protection Concept
• All data remains in original storage. Data for publications can be extracted based on consent from drivers
Sharing UDRIVE data
Enablers • Re-use of the data agreed in project documents • Remote access method tested in project • Application procedure including Data Protection
Concept Obstacles • Funding for post-project data storage and access not
yet solved • Data protection of PID (necessary!)
Other projects
• Application on PTW and cyclist for H2020
• FOT-NET DATA • Naturalistic Cycling
35
• • January 2014 – December 2016 • Budget 1.8M€
• • Support efficient sharing and re-use of FOT datasets • Develop and promote a framework for sharing data • Build a detailed catalogue of available data and tools • Create international networking platform
Partner Network
Bicycle instrumentation
38
Towards Naturalistic Cycling
More information?
[email protected] http://www.udrive.eu