PROSPECT - PROactive Safety for PEdestrians and CyclisTs
Advancing active safety towards the protection of
Vulnerable Road Users:
The project PROSPECT (PROactive Safety for PEdestrians
Ilona CIESLIK, Applus IDIADA
• Project Overview
• Research Activities
• Research Results
• Impact & Next Steps
• Accidents involving pedestrians and cyclists are a pending
issue: they account for 30% of road fatalities globally.
• First generation of Autonomous Emergency Braking Systems
that avoid and mitigate VRU accidents are in the market.
• Euro NCAP roadmaps:
• AEB-Pedestrian systems in 2016.
• AEB-Cyclist systems in 2018.
• More complex scenarios in 2020.
Source: Robert Bosch, 2015
PROSPECT aims to significantly improve the
effectiveness of active VRU safety
systems compared to those currently on the
(1) expanding scope of scenarios
addressed by the systems and
(2) improving overall system performance
based on sensors fusion and VRU modeling.
The emphasis is on two groups with large
shares of fatalities: cyclists and
PROSPECT PROactive Safety for Pedestrians and CyclisTs
EC Call H2020 Two Stages, MG 3.4 2014 “Traffic safety analysis and
integrated approach towards the safety of Vulnerable Road Users”
Type of action Project budget EU Funding Project
€ 6.931.978 € 6.931.978 1st of May 2015
9 EU countries
§ Vehicle manufacturers: Audi, BMW, Volvo, TME, Daimler
§ Tier-1 suppliers: Bosch, Continental
§ SMEs: 4activeSystems
§ Research centres & Test labs: Applus IDIADA, BASt, VTI, TNO,
§ Academia: Univ. of Nottingham, Univ. of Budapest, Univ. of Amsterdam,
Use cases &
sensing & modelling
Actuation & control
Test protocol &
1 2 3 4
The PROSPECT project technical approach
- Kick off
European OEMs: Audi, BMW,
DAIMLER, TME, VCC -
further work on the next
- Draft version of
and user needs
- Project internal
workshop with all
of all system
demo cars, Audi
Active Safety (AEB), testing
Addressed Euro NCAP
scenarios coming in 2020-2024
evaluation phase &
presentation at final
Research Activity 1- Study of relevant VRU scenarios
1 Better understanding of relevant VRU scenarios
• Macro statistical and in-depth accident analyses:
• National statistics from specific countries
• Detailed understanding from GIDAS (DE) & IGLAD (CZ, ES, FR, IT, FR
• CARE analysis for weighting to EU level
• Definition of traffic conditions and user expectations:
• Naturalistic urban observations with large number of VRUs.
• Hotspots monitoring in different EU cities.
Research Activity 2 - AEB Specification
Representative parameters to
derive use cases
Car-to-VRU Use Cases:
• Crossing scenarios
• Longitudinal scenarios
• Turning scenarios
Simulation of different sensor types Visualization of PROSPECT use case
Gesture analysis and description:
• Ask for yielding, reprimand, give the
way, cyclist turn indication, thanks.
October 2016: 1st Year
D3.1 The addressed VRU scenarios within PROSPECT and associated test catalogue 01-may-16
D3.2 Specification of the PROSPECT demonstrators 01-may-16
D2.1 Accident Analysis, Naturalistic Driving Studies and Project Implications 01-nov-16
D2.2 Specification of User Needs and Functional Requirements 01-nov-16
D7.4 Test protocol as a proposal for consumer testing 01-nov-16
• List of project deliverables
• List of project publications
International Cycling Safety Conference 2016, Bologna, Italy “Car-to-cyclist accidents from the
Drivers Point Of View.”
International Cycling Safety Conference 2016, Bologna, Italy “Understanding the cues and
characteristics that indicate and affect a cyclist’s future path: A focus group study conducted in
the UK and Netherlands.”
DASIP2016 “Hardware Acceleration Of Maximum-Likelihood Angle Estimation For Automotive
Research Activity 3 - Advanced VRU sensing (radar & video systems)
Radar-based obstacle detection:
• Higher field of view
• High resolution RADAR with µ-Doppler
• The further development towards safety features:
• VRU classification capabilities needed.
• More accurate position /tracking to select the best mitigation /avoidance
Obstacle detection with radar systems: a)
Generation of µ-Doppler signatures for feature
Video-based obstacle detection:
• The stixel world & Box detection methods
• Advanced machine learning techniques
Semantic stixel representation – the image is
segmented into drivable road, sky, and vertical
"sticks" (Schneider, 2016)
Research Activity 4 - Advanced system control strategies
• Automatic intervention is made transparent and understandable for the driver.
• Transition between driver warning & automated vehicle action are carefully designed:
• Accident avoidance by combined steering and/or braking.
• High dynamic actuators and torque vectoring by braking.
• Challenge: estimation of the criticality in the road scene.
Road participant sand situation analysis with intervention strategy
Research Activity 5 - Integration in demonstrators
• Demonstrators implementing Use Cases:
• Driving simulator.
• Realistic pedestrian and cyclist dummies including platform propulsion system.
• Sensor fusion – overcoming challenges:
• Fusion between same /different
• Calibration, overlapping, sensor
• Definition of interfaces and property
• Analytical comparison different fusion
• List of project deliverables
• List of project publications
The ESV conference 2017, Detroit, US:
- “Car-to-cyclist crashes in Europe and derivation of use cases as basis for test scenarios of next generation
advanced driver assistance systems – results from PROSPECT.”
- “Advancing active safety towards the projection of vulnerable road users: the PROSPECT project.”
“Naturalistic observation to investigate conflicts between drivers and VRUs in the PROSPECT project.”
- “Next-generation Active Safety and Testing: The Horizon 2020 Project PROSPECT.”
The Road Safety and Simulation conference, The Hague, The Netherlands, October 2017 “Exploring the
relationship between false alarms and driver acceptance of a pedestrian alert system during simulated driving.”
The IEEE Intelligent Vehicles Conference, USA “Using Road Topology to Improve Cyclist Path Prediction. ”
October 2016: 2nd Year
Research Activity 6 - Testing and Validation
• Testing in realistic traffic scenarios:
• Real world scenarios to be reproduced in controlled environments.
• Testing in driving simulator & warning with driver
• User acceptance tests:
• Influence of false warnings and incorrect system interventions.
• Predictive model of acceptance.
• Test methodology, procedures & tools to be proposed to
• Unjustified system interventions.
• Intervention performance tests considering evasive actions.
object are symmetric
Research results & examples
• Tangible & not tangible results:
• The generation of state-of-the-art knowledge but as well technical innovations.
Research results & examples - demo-vehicles
• Demonstrator Vehicle II
• High-resolution stereo camera system for 75° coverage
• Fusion with additional short-range RADAR sensor
• Automatic steering and braking
• Use cases: cyclists and pedestrians, crossing and longitudinal
where car can have high speed and early detection needed.
a) Calibrated and synchronized stereo camera and lidar system.
b) Sensor setup consisting of one front facing stereo camera (~60m, 75° ) and two side-
oriented cameras covering a horizontal FOV of roughly 210°.
c) The addressed use cases
Research results & examples - Mobile driving simulator
• Mobile driving simulator:
• To present and evaluate the results of PROSPECT in a realistic setting
applying a real car as a mock-up.
• To demonstrate the circumstances of car-to-cyclist-accidents (the
• Better understanding of possible reasons why car drivers fail to handle such
Audi Mobile driving simulator presentation during PROSPECT technical meeting at BASt.
• Dummies with additional degrees-of-freedom allow intention detection
Research results & examples - test tools and targets
a) Pedestrian dummy full stop and rotating head towards approaching car b) Pedaling cyclist
dummy with rotating wheels.
• Test results, methodologies and assessment protocols for Euro NCAP AEB
• IDIADA, TNO and BAST are members of Euro NCAP Working Group on
• For PROSPECT important will be the Euro NCAP roadmap for 2020 - 2024
which will include requirements for e.g. steering intervention & cross-junction
AEB systems that need to address VRU
Research results & examples - Protocols for Euro NCAP
• PROSPECT's goal: development of novel active safety features to
prevent accidents with VRU.
• Accident analysis and PROSPECT use cases enable the development
of improved VRU sensing technologies and vehicle control
• Prototypes currently in development will be tested in the first half of
• Test methodologies (test scenarios and tools) will be delivered to Euro
• Methodology for benefit analysis:
• Improvement for traffic safety in terms of saved lives or serious
• Assessment of the combined effect of active and passive safety
measures (integrated safety).
• Final demonstration to the public during the final PROSPECT event in
Conclusions & Next steps
Dissemination activities: Conferences May – November 2017
29 Oct – 2 Nov 2017
11-12 May 2017
5-8 June 2017
11-14 June 2017
4-6 September 2017
21-22 September 2016
8-13 October 2017
17-19 October 2017
10 Oct 2017
21 Sep 2017
6-8 September 2016
14-16 Nov 2017
7th International Cycling Safety Conference
& PROSPECT final event
This project has received funding from the European Commission’s
Innovation and Networks Executive Agency, under the frame of Horizon
2020 programme, with grant agreement nº 634149
9 EU countries
5 research centres
Andrés Aparicio firstname.lastname@example.org
Laura Sanz email@example.com
Ilona Cieslik firstname.lastname@example.org
Tel. +34 977 166 717
Fax + 34 977 166 036
• Visit our website www.prospect-project.eu
R&D Project Engineer