Car-to-cyclist Crashes in Europe and Derivation of Use Cases as Basis for Test Scenarios of Next Generation Advanced Driver Assistance Systems Marcus Wisch (BASt*) M Lerner (BASt), J Kovaceva (Chalmers), A Bálint (Chalmers), I Gohl (Uni BW), A Schneider (Audi), J Juhász (BME), M Lindman (Volvo) 25th ESV Conference, Detroit, 06 June 2017 * Federal Highway Research Institute, Germany
Objectives PROSPECT: Improvement of the effectiveness of active Vulnerable Road User (VRU) safety systems compared to those currently on the market This study: Up-to-date crash data analysis of crashes in Europe between passenger cars and cyclists Identification of Accident Scenarios Derivation of the most important Use Cases as basis for the development of Test Scenarios for Advanced Driver Assistance Systems Marcus Wisch 2/16
Method Available datasets and data query Accident statistics have been analyzed towards the injury severity of cyclists in crashes with passenger cars: Killed and seriously injured (KSI) Slightly, seriously injured and killed Focus on crashes with two crash participants Country / Region High-level Crash data In-Depth Crash data Europe CARE - Sweden STRADA Volvo Cars Cyclist Accident Database Germany DESTATIS GIDAS Hungary KSH Pest County Marcus Wisch 3/16
Results Crashes with cyclists (Germany, 2011-2014) Killed and seriously injured cyclists by crash opponent and age group (single crashes or crashes with two participants involved). Car-to-cyclist crashes have highest priority. Mid-aged and older cyclists suffer most often from high injury severities. Marcus Wisch 4/16
Results Faults in car-to-cyclist crashes (Hungary, 2011-2014) Cyclists Car drivers Priority rule violation was observed most often for car drivers as well as for cyclists. Marcus Wisch 5/16
Definitions Accident Scenarios: Movements of crash participants prior to the crash, described by the Accident types ; course of the road, light conditions, weather conditions, view obstruction etc. (I) Car straight on, Cyclist from near-side Exemplary pictograms (II) Car straight on, Cyclist from far-side (III) Car turns (either near- or far-side) (IV) Car and cyclist in longitudinal traffic Marcus Wisch 6/16
Accident Scenarios Results from Germany, Hungary and Sweden (I) Car straight on, Cyclist from near-side (II) Car straight on, Cyclist from far-side (III) Car turns (either near- or far-side) (IV) Car and cyclist in longitudinal traffic Results varied a lot between the countries, in particular regarding Accident Scenario Groups III and IV. (V) Others I II III IV V Parking Total Germany KSI 27% 17% 27% 10% 19% 1% 100% Killed 25% 23% 12% 25% 14% 0% 100% Serious 27% 16% 27% 10% 19% 1% 100% Slight 27% 14% 28% 9% 20% 2% 100% Hungary KSI 42% 6% 29% 23% 0% 100% Killed 33% 1% 64% 2% 0% 100% Serious 42% 7% 26% 24% 0% 100% Slight 42% 8% 24% 26% 0% 100% Sweden KSI 28% 33% 13% 17% 10% 101% Killed 23% 27% 3% 40% 8% 101% Serious 29% 34% 15% 11% 11% 100% Slight 34% 33% 12% 7% 13% 99% Marcus Wisch 7/16
Method Five-Stage-Approach to derive Use Cases Car-to-cyclist crashes in urban areas (GIDAS, 2000-2013, N=4,272) Crashes with frequent coded accident types (>1%) were selected Adding of more detailed information about the road layout, right of way, as well as manoeuvre intention of the car driver Crashes with similar parameters were assigned to the same Use Case Marcus Wisch 8/16
Method Identified Use Cases were projected to all police-reported crashes Ranking considered the frequency and the cyclist s injury outcome (approach developed in the EC funded FP7 project ASSESS): Marcus Wisch 9/16
Results Identified Use Cases Distinction of crash injury outcome: - Part I: - Seriously injured or killed cyclist (KSI), N = 515 29 Use Cases - Part II: - Slightly, seriously injured or killed cyclist, N = 2,669 35 Use Cases Use Cases differ in both groups! However, main Use Cases are the same! Use Cases for part II accounted for 62% of all car-to-cyclist crashes (based on N=4,272 crashes). Marcus Wisch 10/16
Results Ranking of Use Cases Ranking for slightly, seriously and fatally injured cyclists (part II) Ranking 1 2 3 4 5 6 7 8 9 10 The first ten Use Cases account for 36% of all car-to-cyclist crashes (based on N=4,272 crashes) Crossing scenarios play a predominant role (compared to longitudinal traffic scenarios) Marcus Wisch 11/16
Results Main findings road traffic regulations 1) Drivers collided more often with cyclists from the near-side when the cyclist violated road traffic regulations or behaved unexpectedly. cyclist violated road traffic regulations - cyclist from near-side driver violated road traffic regulations - cyclist from far-side Marcus Wisch 12/16
Results Main findings right-of-way, cyclist from farside 2) When driver had no right-of-way, collisions from the far-side had a higher relevance compared to situations, in which the driver had right-of-way: driver violated road traffic regulations cyclist violated road traffic regulations Marcus Wisch 13/16
Discussion Cyclist s riding direction: VS. Car driver s manoeuvre intention: Cyclist on the bicycle lane Riding against road driving direction is usually not compliant with road traffic regulations Drivers inappropriate expectations lead to improper allocation of attention Drivers fail to look to the nearside Sometimes view obstructions (26% vs. 12%) VS. Cyclist from farside Allocation of attention for cyclist s riding direction is correct Drivers look but fail to see One-fourth of these crashes in dark light conditions (28% vs. 23%) Marcus Wisch 14/16
Conclusions General crash data analyses confirmed: Older cyclists have the highest risk to get fatally injured; Male cyclists were injured more often than females; Higher injury severities on rural roads; and Most crashes in fine weather and daylight conditions. Accident Scenarios: KSI cyclists - Results for Germany, Hungary and Sweden were similar regarding Accident Scenarios (I) and (II); around 42%-52% of all casualties were assigned to these scenarios. However, results varied a lot between the considered countries for Scenarios (III) and (IV). Use Cases: Collisions often because of: Violation of traffic regulations (both, car drivers and cyclists) Drivers failed to look due to inappropriate / wrong expectations Drivers looked but failed to see/react due to cyclist s appearance Marcus Wisch 15/16
Partners See also PROSPECT Deliverable 2.1 Accident Analysis, Naturalistic Driving Studies and Project Implications. For further information: Andrés Aparicio (PROSPECT Coordinator) ADAS Product Manager, Electronic Chassis Control Systems aaparicio@idiada.com This project has received funding from the European Union s Horizon 2020 research and innovation programme under grant agreement No 634149 www.prospect-project.eu
Appendix Marcus Wisch 16/17