Traffic Injury Prevention (2014) 15, S183–S189 Published with license by Taylor & Francis ISSN: 1538-9588 print / 1538-957X online DOI: 10.1080/15389588.2014.928930
A Real-Life Based Evaluation Method of Deployable Vulnerable Road User Protection Systems RIKARD FREDRIKSSON1, MIKAEL DAHLGREN2, MARGRIET VAN SCHIJNDEL3, STEFANIE DE HAIR3, and SJEF VAN MONTFORT3 1
Autoliv Research, V˚arg˚arda, Sweden Autoliv Sverige, V˚arg˚arda, Sweden 3 TNO, Helmond, The Netherlands 2
Received 18 March 2014, Accepted 24 May 2014
Objective: The aim of this study was to develop a real-life-based evaluation method, incorporating vulnerable road user (VRU) full-body loading to a vehicle with a deployable protection system in relevant test setups, and use this method to evaluate a prototype pedestrian and cyclist protection system. Methods: Based on accident data from severe crashes, the most common scenarios were selected and developed into 5 test setups, 2 for pedestrians and 3 for bicyclists. The Polar II pedestrian anthropomorphic test device was used, either standing or on a standard bicycle. These test setups could then be used to evaluate real-life performance of a prototype protection system, regarding both positioning and protection, for vulnerable road users. The protection system consisted of an active hood and a windshield airbag and was mounted on a large passenger car with a conventional hood-type front end. Injury evaluation criteria were selected for head, neck, and chest loading derived from occupant frontal and side impact test methods. Results: The protection system managed to be fully deployed, obtaining the intended position in time—that is, before VRU body contact—in all test setups, and head protection potential was not negatively influenced by the preceding thoracic impact. Head loading resulted in head injury criterion (HIC) values ranging up to 4400 for the standard car, and all HIC values were below 650 with the protection system. The risk of severe (Abbreviated Injury Scale [AIS] 3+) head injury decreased from 85% to 100% in 3 test setups (mainly to the windscreen frame), to less than a 20% risk in all setups. In general, there were larger differences between structures impacted than between the pedestrian and cyclist setup. Neck loading was maintained at an acceptable level or was slightly decreased by the protection system, and chest loading was decreased from high values in 2 test setups in which the cyclist was impacted laterally with chest impact mainly to the hood area. Conclusions: A test method was developed to evaluate a more real-life-based test condition, as a complement to current component test methods. Being real-life based, including full-body loading, it is suggested as a complementary test method to the more simplified legal and rating component tests. Together these test methods will provide a more thorough evaluation of a protection system. The evaluated protection system performed well regarding both positioning and protection, indicating a capability to obtain the intended position in time with the potential to prevent the most common severe upper-body injuries of a pedestrian or cyclist in typical real-life accidents, without introducing negative side effects. Keywords: pedestrian, cyclist, vulnerable road user, VRU airbag, active hood, full-body test
Introduction Worldwide it is estimated that over 500,000 pedestrians and bicyclists are killed annually in road traffic (Naci et al. 2009). Virtually all road pedestrian fatalities and a majority of the cyclist road fatalities are caused by crashes with vehicles (Statens © Rikard Fredriksson, Mikael Dahlgren, Margriet van Schijn-
del, Stefanie de Hair, and Sjef van Montfort Associate Editor Joel Stitzel oversaw the review of this article Address correspondence to Rikard Fredriksson, Autoliv Research, Wallentinsv¨agen 22, 447 83 V˚arg˚arda, Sweden. E-mail: [email protected] Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/gcpi.
¨ Kommunikationsanalys 2009). Steps have been Institut for taken in Europe and Asia to legislate minimum pedestrian protection performance for passenger cars, and consumer organizations rate pedestrian protection performance of cars, though for bicyclist protection such activities have not been introduced. European and U.S. accident data show that most common injury combinations for severely injured are head-towindshield and leg-to-bumper and for fatal accidents headto-windshield followed by chest-to-windshield and hood areas (Fredriksson et al. 2010; Longhitano et al. 2005; Mallory et al. 2012). Cars can be designed to mitigate these injuries by increasing the energy absorption distance of the front end, which has been proven effective (Strandroth et al. 2011). But to reach a high protection potential, approximately 100 mm of deformation distance is needed for the front-end up to the
184 windshield area. Though this appears feasible for the bumper area, another solution for the hood and windshield area is to equip the car with deployable systems, such as active hoods and windshield airbags. In larger European cities, bicycle transportation is increasing (Thiemann-Linden 2010), likely due to congestion, fuel prices, and an increasing awareness of its health benefits. Pedestrians and bicyclists already make up roughly half the traffic fatalities in urban areas in the European Union (ERSO 2012), risking fatalities to increase with increased bicycle use. Accident data show that car impact speeds in bicyclist crashes are similar to pedestrian accidents. Though pedestrians are mostly impacted from the side, bicyclists are still most often impacted from the side but to a greater extent impacted from the rear. Bicyclist head impacts also tend to be located higher on the car than pedestrian head impacts (Fredriksson et al. 2012; Fredriksson and Ros´en 2012b; Maki et al. 2003). Legal and rating test methods exist for passive pedestrian protection but are lacking for passive bicyclist-to-car crash protection. Existing pedestrian test methods are based on component impactor tests with isolated and simplified legform and headform tests, which have their advantages whereby they can be used to test all relevant areas of the car where a pedestrian impact is likely to occur with high repeatability. If a full-body anthropometric test device (ATD) were to be used, a large range of ATD sizes and test positions would be required to cover all relevant impact points and with a lower repeatability. But a component test method, with a headform, upper legform, and legform does not consider the influence of other body parts, such as the neck or chest. In addition, when an advanced protection system, such as an active hood and airbag is developed, the headform cannot evaluate potential problems of positioning or influence of body region interaction; for example, shoulder impact influence on head protection. It is essential to develop a test method to assess real-life performance of such protection systems, as a complement to the current subsystem component test methods. Therefore, the aim of this study was to develop a real-lifebased evaluation method incorporating a full-body ATD and a vehicle with a protection system in relevant test setups and evaluate a prototype pedestrian and cyclist protection system.
Method Test Setup The Polar II 50th percentile male ATD (Akiyama et al. 1999, 2001, 2008) was used in 2 pedestrian configurations (pedestrian hit laterally on the vehicle far side and centerline) and 3 bicyclist configurations (bicyclist hit laterally on vehicle far side and near side and bicyclist rear impact with vehicle right offset impact); see Figure 1. These configurations were chosen because they were the most common in severe and fatal pedestrian and cyclist crashes with passenger cars (Fredriksson et al. 2012; Fredriksson and Ros´en 2012). The bicyclist ATD was positioned with the right leg (impacting leg) positioned in the lowest pedal position. The saddle height was set at 100 cm and
Fredriksson et al.
Fig. 1. Pedestrian test setups and dummy position (top) and bicyclist setups/position (below).
the steering bar center at 110 cm, based on the average heights in a Dutch study. The pedestrian ATD was positioned with the impacted leg (right leg) rearwards and with the hands tied in front. A large family car, Volvo V70 model year 2011 (see Figure 2), impacted the vulnerable road user at 40 km/h in all test setups, and the bicyclist/bicycle had an impact speed of 15 km/h in the side impact setups, based on the most common impact speeds in severe VRU-to-car crashes (Fredriksson and Ros´en 2012). The ATD was released 10 ms prior to first impact, and car brakes were applied 250 ms after first impact (the impact is over at approximately 170–220 ms). Further details of the test setup are described in van Schijndel et al. (2012).
Protection System A prototype protection system, developed to protect both pedestrians and bicyclists, was used in the study. It consisted
Vulnerable Road User Protection Systems
185
Fig. 2. Car model used in tests and protection system, consisting of active hood and windshield airbag.
of 2 hood lifters and a windshield airbag. The piston hood lifters lifted the rear corners of the hood ∼100 mm in approximately 50 ms. The airbag of approximately 200-L volume, was positioned in approximately 90 ms and designed to protect in the lower windshield/instrument panel area and in the A-pillar area up to a wraparound distance of 2,500 mm (see Figure 2). Contact sensors were mounted in the bumper area but only to collect data for evaluation. Due to the uncertainty of activation for the bicyclist situation, the protection system was activated by a laboratory sensor indicating contact between ATD or bicycle and the car, with a normal delay used in pedestrian sensor algorithms. Injury Evaluation No injury assessment reference values (IARVs) exist for the Polar II ATD. Pedestrian impact is a complex impact, with possibilities of both frontal and lateral, or combined, loading. Therefore, a review of relevant existing IARVs was performed for both frontal and side impact crash dummies. It is important to point out that it is not possible to use an IARV directly from a different ATD without validation for the specific ATD, in this case the Polar II. But the intention of this study was to find rough values to enable comparisons of different test setups with 2 protection levels for different body regions and multiple IARVs. When several IARVs existed for the same load case, the most conservative value was chosen. Neck injury criteria have been developed only for frontal impacts; side impact regulations do not assess neck loading. Therefore it was essential to find some limits for lateral loading. For the head, Head Injury Criterion 15 ms (HIC15 ) was chosen, which assesses combined linear loading for all 3 axes, and the IARV value of HIC 700 based on the most stringent regulation (FMVSS 208, Hybrid III; NHTSA 2012). For the upper neck shear force it was decided to use
Fig. 3. Right side views of tests with protection system at time instance when body is loading protection system prior to head impact.
the same IARV in lateral loading, 3,100 N, as regulated for frontal (ECE R94, Hybrid III). For neck tension and compression force the most conservative values were chosen (3,300 and 4,000 N; ECE R94, FMVSS 208; NHTSA 2012). In bending moments there are different levels for flexion (190 Nm) and extension (57 Nm; ECE R94, Hybrid III), and for lateral loading we used the intermediate value (123.5 Nm), the same rationale as used by a working group
186
Fredriksson et al.
on side airbag out-of-position loading to children (International Institute for Highway Safety 2003). For Nij , the existing limit of 1.0 from frontal requirements was chosen (FMVSS 208, Hybrid III; NHTSA 2012). Nij is a combined neck injury criterion, where “ij” stands for 4 different injury mechanisms: te, tf, ce, and cf (t = tension, c = compression, e = extension, f = flexion). Finally, for lateral chest deflection the most conservative value (42 mm) was chosen (ECE R94, EuroSID). See Table A1 (online supplement) for an overview. Finally, the risk of sustaining a severe (Abbreviated Injury Scale [AIS] 3+) head injury was evaluated using a risk function developed by the NHTSA (1995). Guidelines for the Evaluation Method of Deployable Protection Systems for Vulnerable Road Users To summarize, we propose the following guidelines for evaluation of deployable protection systems for vulnerable road users: • Use an appropriate anthropometric test device in several test configurations, those most likely to occur in real accidents. • Evaluate positioning to ensure that the protection system can deploy as intended, that the intended protection area is covered, and that the positioning is completed prior to first body loading. • Ascertain that the protection performance for the head is not limited by earlier contact of other body parts; for example, shoulder contact compressing airbag and thereby limiting head protection. • Evaluate head, neck, and chest loading. Priority to head loading, but neck and chest loading should be monitored to ensure that these values do not increase with the protection system.
Results Positioning Time, Covered Area The deployable system (airbag and deployable hood) positioned well and in good time before body contact. The deployable hood was in position well before the thorax and shoulder impact, and the airbag was positioned well in time before the concerned body regions, shoulder and head, impacted the airbag. The deployable hood was positioned after 39–48 ms and thoracic impact to hood ranged from 80 to 100 ms. The airbag was positioned after 91–98 ms and shoulder/head impact to airbag ranged from 115 to 195 ms. Thorax or Shoulder Not Limiting Positioning or Head Protection The second control was to evaluate whether any body part limited activation of the airbag. Because the airbag is activated when the hood is in the elevated position, there could be a potential problem of the shoulder of the pedestrian pushing the hood down and limiting the airbag positioning. This was
Fig. 4. HIC values normalized to injury assessment reference value (IARV HIC = 700), for reference and protection system for all 5 test setups (top), and the same values calculated into risk of AIS3 + head injury (bottom).
studied in all setups, and in all tests the airbag positioned as expected (Figure 3). In one test setup, with the cyclist impacted from the rear, there was a partial airbag strike-though by the head to the A-pillar, which resulted in elevated injury levels for the head and neck (see following section). Therefore, the airbag was redesigned with a wider coverage area on the A-pillar. Because it was estimated that this did not influence test results from the other setups only this test setup was retested with the new airbag design. Results from both tests will be presented. Injury Values Head Peak head acceleration values ranged from 104 to 309 g for the reference tests and between 55 and 245 g for the first round of bag tests (see Figure A1, online supplement). As mentioned above, the airbag was redesigned to avoid partial strike-through, and with the new design the peak values ranged from 55 to 103 g. The HIC values decreased similarly from a range of 300–4,422 for the reference tests to 286–635 for the updated airbag tests (Figure 4, top, and Table A2, online supplement). For the same setups, HIC reductions were achieved with the protection system for all but one test setup; see Figure 4. This was the pedestrian centerline test, with a head impact to the rear part of the hood. In this case, both the reference value
Vulnerable Road User Protection Systems
187
Fig. 6. Chest values normalized to injury assessment reference value (IARV chest deflection = 42 mm) for all 5 test setups.
side and cyclist–near side, both for the lower rib measured. In the cyclist–far side setup the chest impact was to the hood side area, on top of the fender, and in the cyclist–near side setup the chest impact was to the rear hood area also to the side (see Figure A2, online supplement). No deflection measurement was available for frontal chest deflection, but loading direction to the chest was close to pure lateral or toward the rear of the thorax.
Discussion Fig. 5. Neck values normalized relative to chosen injury assessment reference values, for reference tests (top) and protection system tests (bottom).
and the value for the protection system were low, 300 and 635, and both were below the proposed IARV values. The risk of sustaining a severe head injury was evaluated using risk curves from NHTSA. Three test setups showed high risk for head injury in the reference setup, an approximately 90% risk of AIS 3+ head injury (Figure 4, bottom). These were all head impacts to, or close to, the windshield frame. These risk values were reduced by the protection system to in all cases below 20%. Neck Neck loading was evaluated using upper neck forces, bending moments, and Nij . To enable an overview, the injury assessment values (IAV) were compared to the most conservative injury assessment reference values (IARV) found (see Method section for selection of IARV). In the reference tests one value exceeded the chosen IARVs and 6 values were above 80% of the IARVs (Figure 5, top). For the protection system, no values exceeded the chosen IARV and 2 values were above 80% of the IARVs (Figure 5, bottom). Absolute values are also presented in Table A2. Chest The highest lateral deflection value of the 2 positions measured (ribs 4 and 7) was compared for the reference car and the car equipped with a protection system (Figure 6). High deflection values were measured in 2 test setups, cyclist–far
The highest injury values were recorded in locations around the windshield frame, which is in line with previous accident studies for pedestrians and bicyclists (Fredriksson et al. 2010, 2012; Fredriksson and Ros´en 2012). The protection system showed large injury reductions for the head and chest, 2 areas that dominate in fatal injuries as well as upper body severe injuries; neck injury, which is less frequent in pedestrian and cyclist accidents, was affected less by the protection system. Chest loading measured the highest values for the lower part of the rib cage. This is probably due to the pedestrian/cyclist kinematics where the pelvis hits the hood edge, followed by a bending/rolling of the upper body toward the hood where the lower part of the thorax impacts first, followed by upper thorax, shoulder, and head. Although the legal tests assess only the hood area for head injury protection, we know from accident data that rather few severe or fatal head injuries are caused by the hood area, whereas chest injury from the hood area is more frequent. The active hood may therefore be a protection system more for the thorax than for the head. Finally, a future study should attempt to assess rotational head loading to estimate brain injury because this is an important loading condition causing injury not assessed in this study. In one test the head loading was increased with the protection system, although to a value still below HIC 700, the chosen IARV level. This occurred in the pedestrian centerline setup, where the head impacted the rear part of the hood; all other head impacts were to the airbag-protected area. Hood stiffness varies between the softer areas with only sheet metal and the stiffer areas where the hood is reinforced. If no structure is limiting below, head impact tests usually result in HIC values below IARV levels, however. The low HIC value in the
188 reference test indicates that there was sufficient energy absorption distance underneath the hood to prevent strike-through. Due to difficulty in exactly repeating test setups with the lifted hood, the head impact location was not identical, resulting in a higher head loading, likely due to an impact this time to a more rigid part of the hood structure. It is difficult to draw conclusions on the differences of pedestrian and cyclist protection, because the impact conditions were different (based on real-life data). We know from earlier studies that bicyclists have higher impact locations compared to pedestrians, and this was accounted for in the proposed airbag design with a higher coverage area than current pedestrian airbags. (The VRU airbag in this study covered up to 2,500-mm wraparound distance compared to typically 2,100 mm for pedestrian airbags.) However, when impacting the airbag the protection seems to be similar for pedestrians and bicyclists. The injury value reductions are similar for the 2 groups. As mentioned in the Method section, standard pedestrian contact sensors were mounted to the bumper but were used only for evaluation, not activating the protection system. Analysis of these data was not included in this study. Although it can be questioned how human-like the contact forces to the bumper from the ATD are, it is advised to study this further in the future to enable including a sensor activation evaluation in the test method. Although active safety systems, such as pedestrian warning and autonomous braking systems, are increasing rapidly and are being promoted by rating institutes, crashes will still occur and passive systems will still be important when a crash is unavoidable. Fredriksson and Ros´en (2012) showed that combining active and passive protection systems for pedestrians significantly increases the protection potential compared to the single systems. Though it seems feasible to improve the protection in the bumper area with increased thickness of the energy absorbing structure of the bumper, the hood area is more difficult, with limited possibilities to rearrange the engine compartment to allow for increased energy absorption with today’s low air resistance to improve fuel efficiency. The windshield frame, and especially the A-pillars, is even more challenging to improve, because the A-pillars need a high stiffness combined with minimal sight restriction. Therefore, deployable devices are especially interesting in this area and may play an important role in the future in an integrated active/passive safety system to maximize the protection of vulnerable road users. This study was limited to one impact speed and one size ATD, and further leg position was not varied for the vulnerable road user ATD. When using experimental tests, and in this case costly full-scale tests, there is a need to limit the number of tests. By using the most common accident scenarios for pedestrians and bicyclists, the most common impact speed, and an average size ATD, the authors estimate that this test setup covers the most relevant real-life severe accident conditions. By using experimental tests, we are certain to have the correct protection system properties, which may be questioned in simulations, but it would be beneficial to complement this assessment method with simulations where a larger number of scenarios as well as pedestrian/cyclist stances and leg posi-
Fredriksson et al. tions can be studied and where repeatability is unquestioned. When validated human body models are available, possibilities are presented to study specific injury directly as well; for example, brain injury. Because this test method is limited to one VRU size and thereby the head impact locations tested, but also has a limitation on repeatability, it is not an appropriate single test method to evaluate the protection of a car in a legal or rating test. However, it is suggested as a complementary test method for anyone developing a deployable protection system to realistically evaluate its real-life performance. When introducing protection systems it is important to evaluate whether they work as intended in a real-life setting; that they are deployed in a correct way; that they do not give any negative side effects; and that they provide appropriate protection. These important parameters are focused on in this study to propose a test method that provides a more real-life-based evaluation of deployable protection systems for vulnerable road users. It provides a complementary test method to the current impactor subsystem test method, and if these 2 experimental test methods are combined with numerical simulations a thorough evaluation of the real-life protection of vulnerable road users can be assessed. The protection system evaluated in this study is designed to mitigate pedestrian and cyclist head injury, evaluated normally only with headform impactor tests. There is no current test method to assess neck or chest loading for pedestrians or cyclists. This study proposes guidelines for a real-life-based evaluation method and use this test method to evaluate the system, both regarding positioning as well as protection potential. Being a real-life-based test method, including full-body loading, it is suggested as a complementary test method to the more simplified legal and rating component tests. Together these test methods will provide a more thorough evaluation of a protection system. It was shown that the protection system managed to position in time, before body contact, for all test setups. Further, the system was not prevented to position correctly by the ATD interacting with the system preceding head impact, nor was the head protection performance affected by the preceding shoulder/thorax impact. The protection system decreased the head loading from, in 3 cases, high values to generally low values for all 5 setups. Because neck injury is rare in real-life accidents, focus was set on evaluating whether the protection system in any way increased neck loading. In general no increase in neck loading could be shown. For the chest, the protection system, in this case most likely the hood part, decreased loading, especially to the lower chest. The protection system performed well for both positioning and protection, indicating that it will be able to position in time with the potential to prevent the most common severe upper-body injuries of a pedestrian or cyclist in a typical reallife accident, without introducing negative side effects.
Acknowledgment We thank Honda R&D Americas for the loan of the Polar II anthropometric test device.
Vulnerable Road User Protection Systems Funding This work was performed as part of the SAVECAP project (www.savecap.org) and was funded by the Swedish government program Vinnova FFI, as well as the Dutch Ministry for Infrastructure and Environment.
Supplemental Materials Supplemental data for this article can be accessed on the publisher’s website.
References Akiyama A, Okamoto M, Ito O, Takahashi Y. Verification of thoracic response of the Polar-II pedestrian dummy. Paper presented at: 2008 International Crashworthiness Conference; 2008; Kyoto, Japan. Akiyama A, Okamoto M, Rangarajan N. Development and application of the new pedestrian dummy. Paper presented at: International Technical Conference of Enhanced Safety of Vehicles (ESV); 2001; Amsterdam, The Netherlands. Akiyama A, Yoshida S, Matsuhashi T, et al. Development of Human-like Pedestrian Dummy. Tokyo, Japan: Japanese Society of Automotive Engineers; 1999. European Road Safety Observatory (ERSO). CARE database—Traffic safety basic facts 2012. 2012. Available at http://ec.europa. eu/transport/road safety/specialist/statistics/care reports graphics/ index en.htm. Accessed March 16, 2014. ¨ Fredriksson R, Bylund P-O, Oman M. Fatal vehicle-to-bicyclist crashes in Sweden—an in-depth study of injuries and vehicle sources. Ann Adv Automot Med. 2012;56:25–30. Fredriksson R, Ros´en E. Integrated pedestrian countermeasures— potential of head injury reduction combining passive and active countermeasures. Saf Sci. 2012a;50:400–407. Fredriksson R, Ros´en E. Priorities for bicyclist protection in car impacts—a real life study of severe injuries and car sources. Paper presented at: IRCOBI (International Research Council On the Biomechanics of Impact) Conference; 2012b; Dublin, Ireland. Fredriksson R, Ros´en E, Kullgren A. Priorities of pedestrian protection—a real-life study of severe injuries and car sources. Accid Anal Prev. 2010;42:1672–1681.
189 Insurance Institute for Highway Safety. Recommended Procedures for Evaluating Occupant Injury Risk From Deploying Side Airbags. The Side Airbag Out-of-Position Injury Technical Working Group (A joint project of Alliance, AIAM, AORC, and IIHS); Arlington, VA: IIHS; 2003. Longhitano D, Henary B, Bhalla K, Ivarsson J, Crandall J. Influence of vehicle body type on pedestrian injury distribution. Paper presented at: Society of Automotive Engineers World Congress; 2005; Detroit, MI. Maki T, Kajzer J, Mizuno K, Sekine Y. Comparative analysis of vehicle–bicyclist and vehicle–pedestrian accidents in Japan. Accid Anal Prev. 2003;35:927–940. Mallory A, Fredriksson R, Rosen E, Donnelly B. Pedestrian injuries by source: serious and disabling injuries in US and European cases. Ann Adv Automot Med. 2012;56:13–24. Naci H, Chisholm D, Baker TD. Distribution of road traffic deaths by road user group: a global comparison. Inj Prev. 2009;15(1):55–59. NHTSA. Final Economic Assessment, FMVSS No. 201, Upper Interior Head Protection. Washington, DC: Author; 1995. Docket No. NHTSA-1996–1762 item ID-0003. NHTSA. FMVSS Standard No. 208; Occupant Crash Protection. Washington, DC: US Department of Transportation; 2012. ¨ Kommunikationsanalys. Road Traffic Injuries 2008. Statens Institut for ¨ Ostersund, Sweden: Author; 2009. Strandroth J, Rizzi M, Sternlund S, Lie A, Tingvall C. The correlation between pedestrian injury severity in real-life crashes and Euro NCAP pedestrian test results. Traffic Inj Prev. 2011;12:604–613. Thiemann-Linden J. Bicycle Use Trends in Germany. Berlin, Germany: ¨ UrbanisGerman Institute of Urban Affairs (Deutsches Institut fur tik); 2010. United Nations Economic Commission for Europe. Regulation No. 94, Revision 2, Uniform provisions concerning the approval of vehicles with regard to the protection of the occupants in the event of a frontal collision. Geneva, Switzerland: Author. 2013. E/ECE/324/Rev.1/Add.93/Rev.2E/ECE/TRANS/505/Rev.1/Add. 93/Rev.2 United Nations Economic Commission for Europe. Regulation No. 95, Revision 2, Uniform provisions concerning the approval of vehicles with regard to the protection of the occupants in the event of a lateral collision. Author; 2014. van Schijndel M, de Hair S, Rodarius C, Fredriksson R. Cyclist kinematics in car impacts reconstructed in simulations and full scale testing with Polar dummy. Paper presented at: International Research Council on the Biomechanics of Impact Conference; 2012; Dublin, Ireland.
A Real-Life Based Evaluation Method of Deployable Vulnerable Road User Protection Systems RIKARD FREDRIKSSON1, MIKAEL DAHLGREN2, MARGRIET VAN SCHIJNDEL3, STEFANIE DE HAIR3, and SJEF VAN MONTFORT3 1
Autoliv Research, V˚arg˚arda, Sweden Autoliv Sverige, V˚arg˚arda, Sweden 3 TNO, Helmond, The Netherlands 2
Received 18 March 2014, Accepted 24 May 2014
Objective: The aim of this study was to develop a real-life-based evaluation method, incorporating vulnerable road user (VRU) full-body loading to a vehicle with a deployable protection system in relevant test setups, and use this method to evaluate a prototype pedestrian and cyclist protection system. Methods: Based on accident data from severe crashes, the most common scenarios were selected and developed into 5 test setups, 2 for pedestrians and 3 for bicyclists. The Polar II pedestrian anthropomorphic test device was used, either standing or on a standard bicycle. These test setups could then be used to evaluate real-life performance of a prototype protection system, regarding both positioning and protection, for vulnerable road users. The protection system consisted of an active hood and a windshield airbag and was mounted on a large passenger car with a conventional hood-type front end. Injury evaluation criteria were selected for head, neck, and chest loading derived from occupant frontal and side impact test methods. Results: The protection system managed to be fully deployed, obtaining the intended position in time—that is, before VRU body contact—in all test setups, and head protection potential was not negatively influenced by the preceding thoracic impact. Head loading resulted in head injury criterion (HIC) values ranging up to 4400 for the standard car, and all HIC values were below 650 with the protection system. The risk of severe (Abbreviated Injury Scale [AIS] 3+) head injury decreased from 85% to 100% in 3 test setups (mainly to the windscreen frame), to less than a 20% risk in all setups. In general, there were larger differences between structures impacted than between the pedestrian and cyclist setup. Neck loading was maintained at an acceptable level or was slightly decreased by the protection system, and chest loading was decreased from high values in 2 test setups in which the cyclist was impacted laterally with chest impact mainly to the hood area. Conclusions: A test method was developed to evaluate a more real-life-based test condition, as a complement to current component test methods. Being real-life based, including full-body loading, it is suggested as a complementary test method to the more simplified legal and rating component tests. Together these test methods will provide a more thorough evaluation of a protection system. The evaluated protection system performed well regarding both positioning and protection, indicating a capability to obtain the intended position in time with the potential to prevent the most common severe upper-body injuries of a pedestrian or cyclist in typical real-life accidents, without introducing negative side effects. Keywords: pedestrian, cyclist, vulnerable road user, VRU airbag, active hood, full-body test
Introduction Worldwide it is estimated that over 500,000 pedestrians and bicyclists are killed annually in road traffic (Naci et al. 2009). Virtually all road pedestrian fatalities and a majority of the cyclist road fatalities are caused by crashes with vehicles (Statens © Rikard Fredriksson, Mikael Dahlgren, Margriet van Schijn-
del, Stefanie de Hair, and Sjef van Montfort Associate Editor Joel Stitzel oversaw the review of this article Address correspondence to Rikard Fredriksson, Autoliv Research, Wallentinsv¨agen 22, 447 83 V˚arg˚arda, Sweden. E-mail: [email protected] Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/gcpi.
¨ Kommunikationsanalys 2009). Steps have been Institut for taken in Europe and Asia to legislate minimum pedestrian protection performance for passenger cars, and consumer organizations rate pedestrian protection performance of cars, though for bicyclist protection such activities have not been introduced. European and U.S. accident data show that most common injury combinations for severely injured are head-towindshield and leg-to-bumper and for fatal accidents headto-windshield followed by chest-to-windshield and hood areas (Fredriksson et al. 2010; Longhitano et al. 2005; Mallory et al. 2012). Cars can be designed to mitigate these injuries by increasing the energy absorption distance of the front end, which has been proven effective (Strandroth et al. 2011). But to reach a high protection potential, approximately 100 mm of deformation distance is needed for the front-end up to the
184 windshield area. Though this appears feasible for the bumper area, another solution for the hood and windshield area is to equip the car with deployable systems, such as active hoods and windshield airbags. In larger European cities, bicycle transportation is increasing (Thiemann-Linden 2010), likely due to congestion, fuel prices, and an increasing awareness of its health benefits. Pedestrians and bicyclists already make up roughly half the traffic fatalities in urban areas in the European Union (ERSO 2012), risking fatalities to increase with increased bicycle use. Accident data show that car impact speeds in bicyclist crashes are similar to pedestrian accidents. Though pedestrians are mostly impacted from the side, bicyclists are still most often impacted from the side but to a greater extent impacted from the rear. Bicyclist head impacts also tend to be located higher on the car than pedestrian head impacts (Fredriksson et al. 2012; Fredriksson and Ros´en 2012b; Maki et al. 2003). Legal and rating test methods exist for passive pedestrian protection but are lacking for passive bicyclist-to-car crash protection. Existing pedestrian test methods are based on component impactor tests with isolated and simplified legform and headform tests, which have their advantages whereby they can be used to test all relevant areas of the car where a pedestrian impact is likely to occur with high repeatability. If a full-body anthropometric test device (ATD) were to be used, a large range of ATD sizes and test positions would be required to cover all relevant impact points and with a lower repeatability. But a component test method, with a headform, upper legform, and legform does not consider the influence of other body parts, such as the neck or chest. In addition, when an advanced protection system, such as an active hood and airbag is developed, the headform cannot evaluate potential problems of positioning or influence of body region interaction; for example, shoulder impact influence on head protection. It is essential to develop a test method to assess real-life performance of such protection systems, as a complement to the current subsystem component test methods. Therefore, the aim of this study was to develop a real-lifebased evaluation method incorporating a full-body ATD and a vehicle with a protection system in relevant test setups and evaluate a prototype pedestrian and cyclist protection system.
Method Test Setup The Polar II 50th percentile male ATD (Akiyama et al. 1999, 2001, 2008) was used in 2 pedestrian configurations (pedestrian hit laterally on the vehicle far side and centerline) and 3 bicyclist configurations (bicyclist hit laterally on vehicle far side and near side and bicyclist rear impact with vehicle right offset impact); see Figure 1. These configurations were chosen because they were the most common in severe and fatal pedestrian and cyclist crashes with passenger cars (Fredriksson et al. 2012; Fredriksson and Ros´en 2012). The bicyclist ATD was positioned with the right leg (impacting leg) positioned in the lowest pedal position. The saddle height was set at 100 cm and
Fredriksson et al.
Fig. 1. Pedestrian test setups and dummy position (top) and bicyclist setups/position (below).
the steering bar center at 110 cm, based on the average heights in a Dutch study. The pedestrian ATD was positioned with the impacted leg (right leg) rearwards and with the hands tied in front. A large family car, Volvo V70 model year 2011 (see Figure 2), impacted the vulnerable road user at 40 km/h in all test setups, and the bicyclist/bicycle had an impact speed of 15 km/h in the side impact setups, based on the most common impact speeds in severe VRU-to-car crashes (Fredriksson and Ros´en 2012). The ATD was released 10 ms prior to first impact, and car brakes were applied 250 ms after first impact (the impact is over at approximately 170–220 ms). Further details of the test setup are described in van Schijndel et al. (2012).
Protection System A prototype protection system, developed to protect both pedestrians and bicyclists, was used in the study. It consisted
Vulnerable Road User Protection Systems
185
Fig. 2. Car model used in tests and protection system, consisting of active hood and windshield airbag.
of 2 hood lifters and a windshield airbag. The piston hood lifters lifted the rear corners of the hood ∼100 mm in approximately 50 ms. The airbag of approximately 200-L volume, was positioned in approximately 90 ms and designed to protect in the lower windshield/instrument panel area and in the A-pillar area up to a wraparound distance of 2,500 mm (see Figure 2). Contact sensors were mounted in the bumper area but only to collect data for evaluation. Due to the uncertainty of activation for the bicyclist situation, the protection system was activated by a laboratory sensor indicating contact between ATD or bicycle and the car, with a normal delay used in pedestrian sensor algorithms. Injury Evaluation No injury assessment reference values (IARVs) exist for the Polar II ATD. Pedestrian impact is a complex impact, with possibilities of both frontal and lateral, or combined, loading. Therefore, a review of relevant existing IARVs was performed for both frontal and side impact crash dummies. It is important to point out that it is not possible to use an IARV directly from a different ATD without validation for the specific ATD, in this case the Polar II. But the intention of this study was to find rough values to enable comparisons of different test setups with 2 protection levels for different body regions and multiple IARVs. When several IARVs existed for the same load case, the most conservative value was chosen. Neck injury criteria have been developed only for frontal impacts; side impact regulations do not assess neck loading. Therefore it was essential to find some limits for lateral loading. For the head, Head Injury Criterion 15 ms (HIC15 ) was chosen, which assesses combined linear loading for all 3 axes, and the IARV value of HIC 700 based on the most stringent regulation (FMVSS 208, Hybrid III; NHTSA 2012). For the upper neck shear force it was decided to use
Fig. 3. Right side views of tests with protection system at time instance when body is loading protection system prior to head impact.
the same IARV in lateral loading, 3,100 N, as regulated for frontal (ECE R94, Hybrid III). For neck tension and compression force the most conservative values were chosen (3,300 and 4,000 N; ECE R94, FMVSS 208; NHTSA 2012). In bending moments there are different levels for flexion (190 Nm) and extension (57 Nm; ECE R94, Hybrid III), and for lateral loading we used the intermediate value (123.5 Nm), the same rationale as used by a working group
186
Fredriksson et al.
on side airbag out-of-position loading to children (International Institute for Highway Safety 2003). For Nij , the existing limit of 1.0 from frontal requirements was chosen (FMVSS 208, Hybrid III; NHTSA 2012). Nij is a combined neck injury criterion, where “ij” stands for 4 different injury mechanisms: te, tf, ce, and cf (t = tension, c = compression, e = extension, f = flexion). Finally, for lateral chest deflection the most conservative value (42 mm) was chosen (ECE R94, EuroSID). See Table A1 (online supplement) for an overview. Finally, the risk of sustaining a severe (Abbreviated Injury Scale [AIS] 3+) head injury was evaluated using a risk function developed by the NHTSA (1995). Guidelines for the Evaluation Method of Deployable Protection Systems for Vulnerable Road Users To summarize, we propose the following guidelines for evaluation of deployable protection systems for vulnerable road users: • Use an appropriate anthropometric test device in several test configurations, those most likely to occur in real accidents. • Evaluate positioning to ensure that the protection system can deploy as intended, that the intended protection area is covered, and that the positioning is completed prior to first body loading. • Ascertain that the protection performance for the head is not limited by earlier contact of other body parts; for example, shoulder contact compressing airbag and thereby limiting head protection. • Evaluate head, neck, and chest loading. Priority to head loading, but neck and chest loading should be monitored to ensure that these values do not increase with the protection system.
Results Positioning Time, Covered Area The deployable system (airbag and deployable hood) positioned well and in good time before body contact. The deployable hood was in position well before the thorax and shoulder impact, and the airbag was positioned well in time before the concerned body regions, shoulder and head, impacted the airbag. The deployable hood was positioned after 39–48 ms and thoracic impact to hood ranged from 80 to 100 ms. The airbag was positioned after 91–98 ms and shoulder/head impact to airbag ranged from 115 to 195 ms. Thorax or Shoulder Not Limiting Positioning or Head Protection The second control was to evaluate whether any body part limited activation of the airbag. Because the airbag is activated when the hood is in the elevated position, there could be a potential problem of the shoulder of the pedestrian pushing the hood down and limiting the airbag positioning. This was
Fig. 4. HIC values normalized to injury assessment reference value (IARV HIC = 700), for reference and protection system for all 5 test setups (top), and the same values calculated into risk of AIS3 + head injury (bottom).
studied in all setups, and in all tests the airbag positioned as expected (Figure 3). In one test setup, with the cyclist impacted from the rear, there was a partial airbag strike-though by the head to the A-pillar, which resulted in elevated injury levels for the head and neck (see following section). Therefore, the airbag was redesigned with a wider coverage area on the A-pillar. Because it was estimated that this did not influence test results from the other setups only this test setup was retested with the new airbag design. Results from both tests will be presented. Injury Values Head Peak head acceleration values ranged from 104 to 309 g for the reference tests and between 55 and 245 g for the first round of bag tests (see Figure A1, online supplement). As mentioned above, the airbag was redesigned to avoid partial strike-through, and with the new design the peak values ranged from 55 to 103 g. The HIC values decreased similarly from a range of 300–4,422 for the reference tests to 286–635 for the updated airbag tests (Figure 4, top, and Table A2, online supplement). For the same setups, HIC reductions were achieved with the protection system for all but one test setup; see Figure 4. This was the pedestrian centerline test, with a head impact to the rear part of the hood. In this case, both the reference value
Vulnerable Road User Protection Systems
187
Fig. 6. Chest values normalized to injury assessment reference value (IARV chest deflection = 42 mm) for all 5 test setups.
side and cyclist–near side, both for the lower rib measured. In the cyclist–far side setup the chest impact was to the hood side area, on top of the fender, and in the cyclist–near side setup the chest impact was to the rear hood area also to the side (see Figure A2, online supplement). No deflection measurement was available for frontal chest deflection, but loading direction to the chest was close to pure lateral or toward the rear of the thorax.
Discussion Fig. 5. Neck values normalized relative to chosen injury assessment reference values, for reference tests (top) and protection system tests (bottom).
and the value for the protection system were low, 300 and 635, and both were below the proposed IARV values. The risk of sustaining a severe head injury was evaluated using risk curves from NHTSA. Three test setups showed high risk for head injury in the reference setup, an approximately 90% risk of AIS 3+ head injury (Figure 4, bottom). These were all head impacts to, or close to, the windshield frame. These risk values were reduced by the protection system to in all cases below 20%. Neck Neck loading was evaluated using upper neck forces, bending moments, and Nij . To enable an overview, the injury assessment values (IAV) were compared to the most conservative injury assessment reference values (IARV) found (see Method section for selection of IARV). In the reference tests one value exceeded the chosen IARVs and 6 values were above 80% of the IARVs (Figure 5, top). For the protection system, no values exceeded the chosen IARV and 2 values were above 80% of the IARVs (Figure 5, bottom). Absolute values are also presented in Table A2. Chest The highest lateral deflection value of the 2 positions measured (ribs 4 and 7) was compared for the reference car and the car equipped with a protection system (Figure 6). High deflection values were measured in 2 test setups, cyclist–far
The highest injury values were recorded in locations around the windshield frame, which is in line with previous accident studies for pedestrians and bicyclists (Fredriksson et al. 2010, 2012; Fredriksson and Ros´en 2012). The protection system showed large injury reductions for the head and chest, 2 areas that dominate in fatal injuries as well as upper body severe injuries; neck injury, which is less frequent in pedestrian and cyclist accidents, was affected less by the protection system. Chest loading measured the highest values for the lower part of the rib cage. This is probably due to the pedestrian/cyclist kinematics where the pelvis hits the hood edge, followed by a bending/rolling of the upper body toward the hood where the lower part of the thorax impacts first, followed by upper thorax, shoulder, and head. Although the legal tests assess only the hood area for head injury protection, we know from accident data that rather few severe or fatal head injuries are caused by the hood area, whereas chest injury from the hood area is more frequent. The active hood may therefore be a protection system more for the thorax than for the head. Finally, a future study should attempt to assess rotational head loading to estimate brain injury because this is an important loading condition causing injury not assessed in this study. In one test the head loading was increased with the protection system, although to a value still below HIC 700, the chosen IARV level. This occurred in the pedestrian centerline setup, where the head impacted the rear part of the hood; all other head impacts were to the airbag-protected area. Hood stiffness varies between the softer areas with only sheet metal and the stiffer areas where the hood is reinforced. If no structure is limiting below, head impact tests usually result in HIC values below IARV levels, however. The low HIC value in the
188 reference test indicates that there was sufficient energy absorption distance underneath the hood to prevent strike-through. Due to difficulty in exactly repeating test setups with the lifted hood, the head impact location was not identical, resulting in a higher head loading, likely due to an impact this time to a more rigid part of the hood structure. It is difficult to draw conclusions on the differences of pedestrian and cyclist protection, because the impact conditions were different (based on real-life data). We know from earlier studies that bicyclists have higher impact locations compared to pedestrians, and this was accounted for in the proposed airbag design with a higher coverage area than current pedestrian airbags. (The VRU airbag in this study covered up to 2,500-mm wraparound distance compared to typically 2,100 mm for pedestrian airbags.) However, when impacting the airbag the protection seems to be similar for pedestrians and bicyclists. The injury value reductions are similar for the 2 groups. As mentioned in the Method section, standard pedestrian contact sensors were mounted to the bumper but were used only for evaluation, not activating the protection system. Analysis of these data was not included in this study. Although it can be questioned how human-like the contact forces to the bumper from the ATD are, it is advised to study this further in the future to enable including a sensor activation evaluation in the test method. Although active safety systems, such as pedestrian warning and autonomous braking systems, are increasing rapidly and are being promoted by rating institutes, crashes will still occur and passive systems will still be important when a crash is unavoidable. Fredriksson and Ros´en (2012) showed that combining active and passive protection systems for pedestrians significantly increases the protection potential compared to the single systems. Though it seems feasible to improve the protection in the bumper area with increased thickness of the energy absorbing structure of the bumper, the hood area is more difficult, with limited possibilities to rearrange the engine compartment to allow for increased energy absorption with today’s low air resistance to improve fuel efficiency. The windshield frame, and especially the A-pillars, is even more challenging to improve, because the A-pillars need a high stiffness combined with minimal sight restriction. Therefore, deployable devices are especially interesting in this area and may play an important role in the future in an integrated active/passive safety system to maximize the protection of vulnerable road users. This study was limited to one impact speed and one size ATD, and further leg position was not varied for the vulnerable road user ATD. When using experimental tests, and in this case costly full-scale tests, there is a need to limit the number of tests. By using the most common accident scenarios for pedestrians and bicyclists, the most common impact speed, and an average size ATD, the authors estimate that this test setup covers the most relevant real-life severe accident conditions. By using experimental tests, we are certain to have the correct protection system properties, which may be questioned in simulations, but it would be beneficial to complement this assessment method with simulations where a larger number of scenarios as well as pedestrian/cyclist stances and leg posi-
Fredriksson et al. tions can be studied and where repeatability is unquestioned. When validated human body models are available, possibilities are presented to study specific injury directly as well; for example, brain injury. Because this test method is limited to one VRU size and thereby the head impact locations tested, but also has a limitation on repeatability, it is not an appropriate single test method to evaluate the protection of a car in a legal or rating test. However, it is suggested as a complementary test method for anyone developing a deployable protection system to realistically evaluate its real-life performance. When introducing protection systems it is important to evaluate whether they work as intended in a real-life setting; that they are deployed in a correct way; that they do not give any negative side effects; and that they provide appropriate protection. These important parameters are focused on in this study to propose a test method that provides a more real-life-based evaluation of deployable protection systems for vulnerable road users. It provides a complementary test method to the current impactor subsystem test method, and if these 2 experimental test methods are combined with numerical simulations a thorough evaluation of the real-life protection of vulnerable road users can be assessed. The protection system evaluated in this study is designed to mitigate pedestrian and cyclist head injury, evaluated normally only with headform impactor tests. There is no current test method to assess neck or chest loading for pedestrians or cyclists. This study proposes guidelines for a real-life-based evaluation method and use this test method to evaluate the system, both regarding positioning as well as protection potential. Being a real-life-based test method, including full-body loading, it is suggested as a complementary test method to the more simplified legal and rating component tests. Together these test methods will provide a more thorough evaluation of a protection system. It was shown that the protection system managed to position in time, before body contact, for all test setups. Further, the system was not prevented to position correctly by the ATD interacting with the system preceding head impact, nor was the head protection performance affected by the preceding shoulder/thorax impact. The protection system decreased the head loading from, in 3 cases, high values to generally low values for all 5 setups. Because neck injury is rare in real-life accidents, focus was set on evaluating whether the protection system in any way increased neck loading. In general no increase in neck loading could be shown. For the chest, the protection system, in this case most likely the hood part, decreased loading, especially to the lower chest. The protection system performed well for both positioning and protection, indicating that it will be able to position in time with the potential to prevent the most common severe upper-body injuries of a pedestrian or cyclist in a typical reallife accident, without introducing negative side effects.
Acknowledgment We thank Honda R&D Americas for the loan of the Polar II anthropometric test device.
Vulnerable Road User Protection Systems Funding This work was performed as part of the SAVECAP project (www.savecap.org) and was funded by the Swedish government program Vinnova FFI, as well as the Dutch Ministry for Infrastructure and Environment.
Supplemental Materials Supplemental data for this article can be accessed on the publisher’s website.
References Akiyama A, Okamoto M, Ito O, Takahashi Y. Verification of thoracic response of the Polar-II pedestrian dummy. Paper presented at: 2008 International Crashworthiness Conference; 2008; Kyoto, Japan. Akiyama A, Okamoto M, Rangarajan N. Development and application of the new pedestrian dummy. Paper presented at: International Technical Conference of Enhanced Safety of Vehicles (ESV); 2001; Amsterdam, The Netherlands. Akiyama A, Yoshida S, Matsuhashi T, et al. Development of Human-like Pedestrian Dummy. Tokyo, Japan: Japanese Society of Automotive Engineers; 1999. European Road Safety Observatory (ERSO). CARE database—Traffic safety basic facts 2012. 2012. Available at http://ec.europa. eu/transport/road safety/specialist/statistics/care reports graphics/ index en.htm. Accessed March 16, 2014. ¨ Fredriksson R, Bylund P-O, Oman M. Fatal vehicle-to-bicyclist crashes in Sweden—an in-depth study of injuries and vehicle sources. Ann Adv Automot Med. 2012;56:25–30. Fredriksson R, Ros´en E. Integrated pedestrian countermeasures— potential of head injury reduction combining passive and active countermeasures. Saf Sci. 2012a;50:400–407. Fredriksson R, Ros´en E. Priorities for bicyclist protection in car impacts—a real life study of severe injuries and car sources. Paper presented at: IRCOBI (International Research Council On the Biomechanics of Impact) Conference; 2012b; Dublin, Ireland. Fredriksson R, Ros´en E, Kullgren A. Priorities of pedestrian protection—a real-life study of severe injuries and car sources. Accid Anal Prev. 2010;42:1672–1681.
189 Insurance Institute for Highway Safety. Recommended Procedures for Evaluating Occupant Injury Risk From Deploying Side Airbags. The Side Airbag Out-of-Position Injury Technical Working Group (A joint project of Alliance, AIAM, AORC, and IIHS); Arlington, VA: IIHS; 2003. Longhitano D, Henary B, Bhalla K, Ivarsson J, Crandall J. Influence of vehicle body type on pedestrian injury distribution. Paper presented at: Society of Automotive Engineers World Congress; 2005; Detroit, MI. Maki T, Kajzer J, Mizuno K, Sekine Y. Comparative analysis of vehicle–bicyclist and vehicle–pedestrian accidents in Japan. Accid Anal Prev. 2003;35:927–940. Mallory A, Fredriksson R, Rosen E, Donnelly B. Pedestrian injuries by source: serious and disabling injuries in US and European cases. Ann Adv Automot Med. 2012;56:13–24. Naci H, Chisholm D, Baker TD. Distribution of road traffic deaths by road user group: a global comparison. Inj Prev. 2009;15(1):55–59. NHTSA. Final Economic Assessment, FMVSS No. 201, Upper Interior Head Protection. Washington, DC: Author; 1995. Docket No. NHTSA-1996–1762 item ID-0003. NHTSA. FMVSS Standard No. 208; Occupant Crash Protection. Washington, DC: US Department of Transportation; 2012. ¨ Kommunikationsanalys. Road Traffic Injuries 2008. Statens Institut for ¨ Ostersund, Sweden: Author; 2009. Strandroth J, Rizzi M, Sternlund S, Lie A, Tingvall C. The correlation between pedestrian injury severity in real-life crashes and Euro NCAP pedestrian test results. Traffic Inj Prev. 2011;12:604–613. Thiemann-Linden J. Bicycle Use Trends in Germany. Berlin, Germany: ¨ UrbanisGerman Institute of Urban Affairs (Deutsches Institut fur tik); 2010. United Nations Economic Commission for Europe. Regulation No. 94, Revision 2, Uniform provisions concerning the approval of vehicles with regard to the protection of the occupants in the event of a frontal collision. Geneva, Switzerland: Author. 2013. E/ECE/324/Rev.1/Add.93/Rev.2E/ECE/TRANS/505/Rev.1/Add. 93/Rev.2 United Nations Economic Commission for Europe. Regulation No. 95, Revision 2, Uniform provisions concerning the approval of vehicles with regard to the protection of the occupants in the event of a lateral collision. Author; 2014. van Schijndel M, de Hair S, Rodarius C, Fredriksson R. Cyclist kinematics in car impacts reconstructed in simulations and full scale testing with Polar dummy. Paper presented at: International Research Council on the Biomechanics of Impact Conference; 2012; Dublin, Ireland.
