Skip to main content

Assessment of Integrated Vehicle Safety Systems for improved vehicle safety

Final Report Summary - ASSESS (Assessment of Integrated Vehicle Safety Systems for improved vehicle safety)

Executive summary:

Road safety is a major societal issue. In 2009, more than 35,000 people died on the roads of the European Union, i.e. the equivalent of a medium town, and no fewer than 1,500,000 persons were injured. The cost for society is huge, representing approximately 130 billion Euros in 2009. In view of this the European road safety policy orientations up to 2020 new technologies that have high potential to improve road safety should be promoted. This includes Integrated Safety Systems (ISS) like pre-crash sensing systems with collision warning and / or autonomous vehicle actions. For such systems it is stated that1 'Accelerated deployment and broad market take-up of such safety enhancing applications needs to be supported in order for their full potential to be unleashed'. The ASSESS project responded to this by developing harmonised test and assessment procedures for the evaluation of pre-crash systems.

Project Context and Objectives:

1.1 Project Context
Road safety is a major societal issue. In 2009, more than 35,000 people died on the roads of the European Union, i.e. the equivalent of a medium town, and no fewer than 1,500,000 persons were injured. The cost for society is huge, representing approximately 130 billion Euro in 2009. In view of this the European road safety policy orientations up to 2020 new technologies that have high potential to improve road safety should be promoted. This includes Integrated Safety Systems (ISS) like pre-crash systems with anti-collision warning or autonomous vehicle actions. For such systems it is stated that 'Accelerated deployment and broad market take-up of such safety enhancing applications needs to be supported in order for their full potential to be unleashed'. The ASSESS project is directly responding to this by developing test and assessment procedures for the evaluation of such systems.

1.2 Main Objectives
The overall objective of the ASSESS Project is to enable widespread introduction of ISS by
A) developing required understanding on the evaluation of Integrated Vehicle Safety Systems and
B) implementing these findings in test and assessment procedures that will set targets for optimal systems in terms of occupant protection.

In view of the above the main objectives of the ASSESS project are:
To design and develop an evaluation tool for pre-crash systems in order to give to the OEM's methodologies to test, perform and standardise pre-crash systems.
-To develop harmonised and standardised assessment procedures and related tools for selected integrated safety systems. Procedures will be developed for driver behaviour evaluation, pre-crash system performance evaluation, crash performance evaluation and socio economic assessment.
-To gain acceptance for future implementation of test and assessment tools in regulatory or consumer rating procedures by extensive evaluation and validation.
-To provide policy recommendations for enabling the implementation of the most important technologies
-To provide an overview of legal barriers that obstruct the introduction of integrated vehicle safety systems
-To analyse the potential socio-economic benefits of the selected integrated vehicle safety systems

1.3 Objectives per work package
1.3.1 WP1 - Definition of targets and final verification
-Prepare a project briefing on previous research on accident scenarios suitable for testing integrated safety systems
-Establish scenario definitions based on existing accident databases and Field Operational Tests
-Establish performance requirements for the test protocol for the different test scenarios
1.3.2 WP 2 - Socio economic evaluation and Legal aspects
-Define Integrated socio-economic impact assessment framework
-Application of the framework to Autonomous Brake systems
-Compile permissibility and important legal constraints of Integrated Safety Systems for different countries
-Generate an abstract outline assessment of liabilities related to the general features of Integrated Safety Systems.

1.3.3 WP 3 - Evaluation of behavioural aspects
-Development of a test and assessment methodology to quantify effects or qualify the interaction of the driver/driver behavioural aspects on the performance of Integrated Safety Systems - suitable for system optimisation, regulatory testing and consumer assessment

1.3.4 WP 4 - Pre-crash evaluation
-Development and evaluation of test and assessment methods for the pre-crash performance of Integrated Vehicle Safety Systems, suitable for regulatory testing and consumer assessment.

1.3.5 WP 5 - Crash evaluation
-Evaluate the performance of several testing and simulation activities by using different passive safety tools to evaluate the benefits coming from the activation of pre-crash systems during the crash phase when the impact has not been avoided.
-Analyse of the capacity of the current passive safety tools to evaluate the effect on the occupants' injuries of the pre-crash systems activation when the impact has not been avoided.

Project Results:

2 Main Science and Technological results and Foreground
The ASSESS project started June 1st 2009 and ended December 31st 2012. In the project an overall methodology for the assessment of integrated safety systems was developed. The methodology integrates driver behaviour testing and pre-crash performance testing in line with real world operation of the system. The methodology also incorporates methods for benefit estimation to judge the effectiveness of systems offered to the market. For analysis of the crash phase existing passive safety tools were used to gain insight in effects of pre-crash actions on the performance of the overall safety system during the crash phase.

2.1 WP1 - Definition of targets and final verification
Accident surveys were conducted to identify test scenarios for pre-crash system performance evaluation. Information from previous EU FP projects in the field of advanced vehicle safety systems like PReVENT, APROSYS was analysed and completed with statistical accident surveys to reveal relevant scenarios. During two project workshops the scenarios were transferred to practical scenarios applicable on test tracks. Although ASSESS collected data for cut-in scenarios, junction scenarios and on-coming scenarios current vehicle systems cannot handle these. For reasons of technology readiness of sensing systems it was decided to focus on rear-end impact scenarios.

2.1.1 Objectives
The first objective was to define casualty relevant accident scenarios to develop test scenarios for active safety system testing based on accident types which currently result in the greatest injury outcome, measured by a combination of casualty severity and casualty frequency.

The second objective was to quantify of the robustness of the test procedures developed in the project with respect to repeatability and reproducibility. This part was performed in close cooperation with Work Package 4 and is presented in respective Chapter.

The third objective was to develop an assessment procedure based on the real world accident population to provide an evaluation of Integrated Vehicle Safety Systems (IVSS).

2.1.2 Accident analysis and test scenario definition

A literature review was carried out to examine how relevant scenarios had been developed by previous projects. The review of previous projects provided a large overview of activities concerning the research in terms of integrated safety. However, the projects, including accident analysis, generally did not define detailed scenarios. Besides the literature review an accident analysis was completed for a range of accident databases to define preliminary accident scenarios. The databases used including those which were nationally representative (STATS19 and STRADA) and in-depth sources which provided more detailed parameters to characterise the accident type (GIDAS and OTS). A common analysis method was developed in order to compare the data from these different sources. While this was not totally successful, the majority of the data was aligned in such a way as to allow a comparison between these databases. These findings can be found in Deliverable D1.1 - Preliminary Test Scenarios .

The principle of the accident analysis was that it considered the real world accident problem in terms of which accident types currently result in the greatest injury outcome. Since the project was focusing on pre-crash sensing systems fitted in passenger cars the data selected for analysis was injury accidents which involved at least one passenger car.

The purpose of defining the preliminary accident scenarios was to rank the most frequent and severe accident scenarios on a high level. All injured people in all involved vehicles (including pedestrians) were taken into account since safety systems in one car can prevent injuries in its own car but also in the counterpart in the accident.

The results of the extended national database analysis on general level were presented in Deliverable 1.2 - Specifications for scenario definitions . Note that the original accident classification system (GDV) which the definition in ASSESS is based on does not only include single vehicle accidents in type 1 'driving accident'. A driving accident can also lead to a collision with other road users, hence multiple vehicles are involved. But in many databases this would be coded as accidents in longitudinal traffic because the initial conflict is not known or because of the definition of the accident type in that database.

Accidents in longitudinal traffic with multiple vehicles were ranked as the number one accident type producing the higher casualty injuries based on injury cost. Following this analysis a detailed analysis of in-depth data was conducted. Based on these two analyses Work Package 4 transferred the findings into to test scenarios.
Each of these subgroups include different test set-ups concerning; driving speeds (urban or motorway), braking (normal driving or emergency braking), offset (no or 50%) and driver reactions (no reaction, slow reaction or fast reaction). The full test matrix for pre-crash system performance evaluation can be found in Deliverable 4.1 - Action plan pre-crash evaluation , pp. 17-21.

However, the test program was quite ambitious and due to the complexity of active safety system testing the work was primarily focused on the rear-end scenarios which include three major sub groups;

-A1: Slower lead vehicle, the lead vehicle (TV) is moving at constant slower speed than the subject vehicle.
-A2: Decelerating lead vehicle, the lead vehicle (TV) is decelerating until stopped.
-A3: Stopped lead vehicle, the lead vehicle (TV) is stopped.


For each of these test scenarios a detailed analysis of the vehicle driving speeds in the accidents were conducted to confirm the relevance of the test speeds. The driving speed data is presented with the percentile-distributions and the analysis results show for all percentiles comparable subject vehicle driving speeds between both databases. For the 75th percentile the speeds can be determined in a range from 50km/h to 65km/h. More details about the analysis can be found in Deliverable D1.2 - Specifications for scenario definitions3.
2.1.3 Over-all system assessment methodology
A new assessment procedure was developed which provides an evaluation of IVSS. The assessment method requires test scenarios that cover the whole speed range that is relevant to rear-end collisions. For the assessment methodology the test protocol proposed by Euro NCAP including incremental test speeds are used. In order to take human-machine interaction (HMI) into account, two tests are planned for each initial condition: one with purely autonomous activation of the pre-crash system and one with driver reaction.

Performance is measured in terms of the percentage of certain injuries that can be avoided with the safety system. In other words, the evaluation is purely injury-based and other, not safety-related factors (such as repair costs) are not taken into account. The proposed method has a scientific basis, but it is also intended to be simple, so that it can be explained to different audiences. The assessment method was discussed in the Harmonisation Platform III and has been presented to EuroNCAP for consideration. The over-all system assessment methodology proposed by the ASSESS project will be presented in short below, see Deliverable D1.4 - Safety impact assessment of integrated vehicle safety systems for further reading.

2.2 WP 2 - Socio economic evaluation and Legal aspects
2.2.1 Introduction
To assess system benefits and costs from a socio-economic perspective a break even methodology was derived. The approach uses accident forecast data from eIMPACT, assumptions for the future vehicle fleet in EU-27 and safety impact analysis based on state of the art injury risk curves and empirical data from ASSESS itself. The method was applied to the systems tested in ASSESS, considering the 2020 - 2030 time horizon. As it was not possible to collect detailed cost data for the systems due to confidentiality reasons, target break even costs were determined indicating the maximum cost that would still provide societal benefits. The break even costs were determined for relevant system configurations namely autonomous braking only versus driver warning in combination with autonomous braking. The study clearly showed the benefits of driver warning and thereby the need for inclusion in future test protocols.

An analysis is legal barriers and issues were performed, considering various countries in different regions of the world. Based on this an outline of methodological tools for the legal assessment of Integrated Vehicle Safety Systems was devised. The tools are meant to contribute to a reduction of product liability risks by design and - at the same time - help to comply with the provisions of the Vienna Convention in terms of controllability.

It turned out that three main areas need to be considered:
-The issue of override-ability of IVSS-triggered / automated interventions,
-The special case of interventions in areas beyond human reaction capabilities and
-The importance of information and/or warning strategies preceding respectively accompanying an automated intervention.

2.2.2 Socio economic evaluation
Since a reliable estimation of system cost is currently very difficult a Break-Even Analysis (BEA) was carried out instead of the classical cost-benefit approach in order to calculate critical safety system cost. Pre-crash functionalities use components of other safety systems such as adaptive cruise control and thus a separation of costs is not possible and connected with some uncertainties due to highly confidential cost information of OEMs. Cost data from other projects like eIMPACT are quite old and do not fit very well to the bandwidth of functionalities of current pre-crash systems and their difference in performance. Since the BEA follows the principles of a classical CBA in order to assess the monetary safety benefit of pre-crash systems, the calculated critical safety system costs represent target break-even costs, which indicate the maximum cost of the system that would still provide societal benefits and therefore would comply with a benefit-cost ratio of more than one.

The main results of the socio-economic assessment can be summarized as follows:
-In the first part of the report background data which are needed for socio-economic assessment are provided. This includes the forecast of fatality figures for the assessment period 2020 and 2030, and a forecast on the passenger car fleet based on ProgTrans data. The estimated fatality trend shows (medium scenario) between 2010 to 2020 a reduction of nearly 35 %, and for the next decade (2020 - 2030) a further fatality reduction of about 9 %. For the European car fleet a growth rate of about 10% between 2010 and 2020 and of about 9% between 2020 and 2030 was estimated.

-Because safety benefits of the system depend on the participation of the equipped cars in traffic, the share of equipped cars was transferred to the share of mileage driven by equipped cars. Since age is a strong predicator of annual car mileage, by using German car market data a linear regression analysis of a functional relationship between (age-dependent) share of car fleet and corresponding (age-dependent) share of mileage driven was specified.
-The core of the safety assessment is a mitigation model from WP1 based on the concept of injury risk curves formalising the relationship between fatality / injury risk and delta-v. In the model a lognormal distribution of accidents over crash severity indicator delta-v is used following US-accident data because similar data at this level of detail are missing for Europe. This accident distribution is used as input for estimation of an impact of the pre-crash system on the accident distribution function. Combined with injury-risk functions then percentage reductions of casualties are predicted. These changes are interpreted as system effectiveness with respect to different reductions of delta-v caused by the pre-crash system. Based on the testing results the mean safety impact of the pre-crash system accounts for an decrease of nearly 55% of fatalities, 29% serious injuries and 20% slight injuries in rear-end accidents.
-For the socio-economic assessment generic pre-crash system variants including warning and autonomous emergency braking were assumed. Based on the results from the safety impact analysis a range of critical safety system costs have been calculated. For the mean scenario with an assumed market penetration of the car fleet with pre-crash systems of 5% in 2020 and 15% in 2030, a reduction of fatality figures ranging from 31 to 34 (2020) and from 51 to 54 (2030) was calculated for the different system variants. The maximum total safety benefit in the mean scenario accounts for nearly 124 million Euro in 2020 and 278 million Euro in 2030 respectively. This results in critical safety system costs of 93 (2020) and 64 (2030) Euro which represent manufacturing cost of the system. Taking the well established 'Factor 3' rule of thumb into account market prices for the pre-crash systems could be in the range of 280 (2020) and 192 (2030) Euro and still be efficient from an overall society point of view.
-The sensitivity of results is addressed at three different levels, starting from parameter changes of the calculation model (market penetration, accident data forecast), then taking a full equipment scenario for 2020 and 2030 into account and finally, enlarging the model to incorporate further safety impacts due to other avoided accidents than only rear-end collisions. For changes of the accident trend and market penetration forecasts the calculated critical safety system costs show depending on the parameter changes a range of 46-121 Euro in 2020 and 34-88 Euro in 2030. Enlargement of the safety benefit assessment with findings from the eIMPACT project show a considerably decrease in total benefits and target break even cost. Critical safety cost amount to 256 Euro in 2020 due to systems effectiveness on other accident types than only rear-end crashes.
-The applied Break-even analysis methodology has proven its applicability to this type of research question. Up-scaling from micro level (testing) to macro level (EU-27 databases for accidents etc.) provides still considerable challenges, especially concerning the granularity of information. Socio-economic assessment makes typically use of averages of variables whereas distributions of variables would be valuable to keep the value added of testing data. Research in this direction would help to solidify the derivation of socio-economic impacts from testing data. Nevertheless, it is demonstrated how the safety assessment tool and the calculated critical system costs can be used to rank pre-crash systems of different functionalities (warning system vs. purely autonomous pre-crash system etc.).

2.2.3 Legal aspects
Integrated Vehicle Safety Systems (IVSS) such as Advanced Emergency Braking Systems (AEBS) may include relevant liability risks. These may arise from system boundaries, from possible malfunctions or from maloperation of the IVSS by the driver (as far as applicable). Product liability is a crucial aspect in this context since a major product liability risk might influence or even hinder the advanced development or the market introduction of a new IVSS.

Starting point of liability / legal considerations were scenarios in which an IVSS does not work in the way it is expected to work from the user´s i.e. the driver´s perspective. Two main scenarios were considered:
-scenario 1: Accidental activation of an automatic emergency braking intervention without obstacle (false-positive intervention)
-scenario 2: Non-activation of an automatic emergency braking intervention in spite of present obstacle

Concerning the above mentioned scenarios the respective legal framework differs within the different automobile markets - it even slightly differs within the EU which has already harmonized certain liability aspects such as product liability since 1985. So task 2.3 concentrates on the present status quo and the legal effects in terms of liabilities concerning IVSS in four different automobile markets: France and Germany as members of the EU, Japan and the U.S. were chosen in this context as important automobile markets:
-France: Recommendation to clearly recall the operating limits and performance restrictions of an IVSS in the advertising documents, instructions for use and at the time of the presentation by the seller of the vehicle thus equipped to the customer; most complete and clearly comprehensible information should be brought to the client (instructions for use, advertising brochures...).
-Germany: Product has to comply with the state-of-the-art in science and technology; controllability in terms of Vienna Convention and RESPONSE 3 Code of Practice should be observed; automated interventions should be overrideable; the user should be instructed comprehensively including system boundaries and warning of blind trust in the system; moreover, no exaggerated expectations (product safety) should be evoked.
-Japan: Recommendation that the manufacturer has to announce the limitation of the system in an appropriate way (similar to the recommendations for French law and for German law).
-U.S. : Responsibility for the safety of IVSS rests with the manufacturer who needs to assure that his products meet what is considered as the technical state of the art (as documented e.g. in US or international industry standards, e.g. SAE or ISO).

These may be condensed in two vital recommendations:
-On the one hand the product should comply with the state-of-the-art in science and technology as much as possible.
-On the other hand the user should be instructed in a most complete and comprehensive way (e.g. by instruction manual, advertising brochures, information in the vehicle) about what the product can do and what it cannot do, i.e. about system boundaries, possible malfunctions, scope of operation. The user should be warned of blind trust in the system; moreover, no exaggerated user expectations should be evoked.

Based on the results of the legal analysis carried out in task 2.3 (see above) task 2.4 proposes a brief and abstract outline of methodological legal tools to allow for the legal assessment of IVSS; the following areas have been identified to be of high relevance for the legal assessment of IVSS:
-Overrideability of IVSS-triggered / automated interventions
From a product liability point of view, as well as with regard to the 1968 Vienna Convention on Road Traffic, it is recommendable to design automated interventions which are triggered by IVSS in a way that allows the driver to override the intervention at any time he wishes to do so.

-Automated interventions in areas beyond human reaction capabilities
Automated interventions in areas which lie beyond human reaction capabilities represent a special case: These interventions will occur at a point in time at which the driver is unable to mitigate or avoid the accident all by himself. The intervention then complies with the will of a carefully acting driver (at least this can legally be presumed under these conditions). For this reason it remains within the legal scope of interpretation to assume these interventions remain in line with the provisions of the Vienna Convention.

-Information and warning strategies
Automated interventions should be preceded and accompanied by information that alerts the driver to the hazard, allowing him to take the appropriate action since this puts the driver's will forward to maximum extent: In case an impending collision is detected, driver's reactions should be called forth by warning strategies which give the opportunity to initiate accident-avoiding braking or steering - or even to override an upcoming intervention if this is appropriate.

2.3 WP 3 - Evaluation of behavioural aspects
2.3.1 Introduction
The objective of ASSESS WP3 was to develop and evaluate a test and assessment methodology to quantify effects or qualify the interaction of the driver/driver behavioural aspects on the performance of Integrated Safety Systems suitable for system optimisation, regulatory testing and consumer assessment.

2.3.2 Story book
The basic idea of the story book is to compare the same traffic scenarios, which differ from the system performances or referent events, with respect to the initiated driver behaviour. Each subject is supposed to experience scenarios described in the story book: TRUEwith (TRUE event with warning), TRUEwithout (TRUE event without warning), REFERENCE, FALSE. Among the 24 possible combinations of the four events not all combinations were implemented considering practical limitations of number of subjects, and to choose only four combinations based on the motivation to minimize any potential expectancy effect by starting with TRUE scenario (TRUEwith or TRUEwithout). Subjects are divided in 4 groups, each group having own order for the four events.

2.3.3 Findings of experimental studies
In order to enable a relevant comparison between TRUEwith and TRUEwithout conditions and see the benefit of a collision warning function, it is necessary to have a scenario which is critical enough in which the driver will have to take an emergency action to avoid the accident. The combination of the intended initial driving conditions (related to speed and relative distance to the target vehicle) and of the secondary task is crucial to generate such critical situation.

One of the difficulties encountered was to control the initial conditions while keeping a natural driving style for the subjects. In driving simulator, the intended initial conditions could be generally achieved by controlling the speed of the leading vehicle. However, in the Toyota's experiments, most of subjects were not comfortable with this driving situation. Since the manoeuvres in the Daimler's experiment correspond to an 'every day' driving situation on the German highway comparable results could not be found. In test track experiments, important variations of speed and distance were observed, resulting in a high rate of invalid subjects for which the situation was not critical enough.

Apart from the scenario's initial conditions, the main parameter which will influence the criticality of the event is the secondary task. Indeed, the visual distraction is expected to be dependent on the combination of the secondary task and of the scenario. As a consequence, any observed benefit of a warning function will be highly dependent on the selected secondary task. The used secondary task did not lead to the expected level of driver distraction. This leads, especially in the Daimler's experiment, where the execution of the secondary task was not monitored by the operator, towards low levels of driver distraction.

The initial test design combined four manoeuvres (TRUEwith, TRUEwithout, FALSE and REFERENCE) per test run and subject in different permutations. Analysis of the results in both driving simulators showed that for TRUE events, the level of expectancy is high in the 4th manoeuvre. After having experienced a critical scenario a few minutes beforehand, most of subjects were prepared to face another critical event, and a significant difference in driver reaction was observed. As a conclusion, such a test design combining more than one TRUE event is less appropriate.

Two KPIs ('Brake Reaction Time' and 'Time To Collision') have been investigated regarding the evaluation of the IVSS's HMI benefit. Both could be suitable to differentiate the effect of a warning function on the driver reaction by comparing their values calculated from manoeuvres with and without an HMI activation. However the differences were not significant for the conducted experiments, explained partially by the limitations of the experimental design.

In both test track and Toyota's driving simulator, the 'stone chipping' sound implemented in the REFERENCE condition was not recognized by the subjects as 'stone chipping', therefore the results have been interpreted as the comparison between 'false alarm' and a 'sound without any meaning'. As for analysis, in both test track and Toyota's driving simulator, the 'steering entropy', selected as an indicator to the mental stress or workload shows that the false alarm may induce more stress than a 'nonsense' sound, not necessarily matching the subjects' answers in the questionnaires. The Daimler' s driving simulator experiments, using a sound system to play the 'stone chipping', shows no significant difference in the 'steering entropy' between the 'stone chipping' and REFERENCE condition, but a significant difference between the events and the phase before.

From the Driving Simulator experiments, there was an attempt to define a 'typical' and 'generic' driver reaction to be applied for further actual vehicle tests in WP4. It has been noted that the driver reaction times are highly dependent on the data selection method, on the tested scenario and on the selected secondary task. The different result obtained from Toyota's and Daimler's experiments illustrates that difficulty to obtain robust reaction times to a warning. Therefore to define a generic driver reaction applicable for different set of scenarios can be hardly done in a robust manner.

However some driver reactions could be quantified from the Toyota's experiments, taking into consideration only the subjects who were effectively distracted at the start of the event. Type of reaction (braking, steering, etc), reaction times to warning and braking profiles where measured. After reviewing this results and other published information, WP3 partners highlighted the following conclusions:
-From literature, a wide range of driver reactions can be observed from different studies.
-Results from the Toyota driving simulator is just one of these various results:
-The reaction times from Toyota Driving Simulator could be considered as a 'worst case' example, only valid for the given scenario ('leading vehicle braking' at 0.7 g) and with the given (highly distracting) secondary task
-The brake force applied will be significantly dependent on the particular brake pedal characteristics of the vehicle.

Based on the interpretation of reaction times from various studies, WP3 partners suggested using the following reaction times as a first step for the use in pre-crash scenarios, but highlighted that further research would be needed to establish a robust driver reaction model:
-25th percentile: 1.2 sec.
-50th percentile: 1.4 sec
-75th percentile: 1.6 sec

2.3.4 Draft test protocol
The findings from the experimental studies are directly incorporated in the draft test protocol for the purposes of optimising and substantiating the story book in sense of a test regulation. The draft test protocol is structured such that it can be applied both in controlled environments and in driving simulations (motion-based). The draft test protocol concentrates on assessing the benefits of the HMI of ISS and calculating the associated key performance indicators required as input for the overall assessment of ISS within ASSESS WP1 and WP4. Compared to the story book the draft test protocol contains two major changes regarding the test design.

The initial test design combined four manoeuvres (TRUEwith, TRUEwithout, FALSE and REFERENCE) per test run and subject in different permutations. The intention was to optimize the number of subjects and to avoid any order or expectancy effects. Analysis of the results in all test facilities showed that for TRUE events, the level of expectancy is high in the 4th manoeuvre. After having experienced a critical scenario a few minutes beforehand, most of subjects were prepared to face another critical event, and a significant difference in driver reaction was observed. As a conclusion, such a test design combining more than one TRUE event is not appropriate. It is recommended that the number of TRUE manoeuvres per subject should be reduced to one.

The findings from the experimental studies had shown the need to rethink the secondary task described in the story book. Three alternative secondary tasks, the peripheral detection task, the arrows task and the auditory continuous memory task were reviewed. The peripheral detection task is a great task to measure cognitive workload during driving. Arrows task have been used in many studies both on-road and into simulators. This visual task uses a little manual component to give the answer. Studies performed earlier have demonstrated that the task is able to create longer visual time off road both in the field and in simulator. The arrow task seems to be appropriate to be used in a test protocol for ASSESS. Moreover, the arrow task presents an advantage to offer the possibility to adapt the level of difficulty for each subject, if needed (change the number of arrows and their orientation).

The development of a draft test protocol, which has been improved with the lessons learnt during the previous testing activities and can be used as reference guideline to study behavioural aspects of ISS like autonomous emergency braking systems or other driving assistance systems

2.3.5 Knowledge transfer
During the Euro NCAP discussions on driver reaction time, the ASSESS results together with other studies showing significantly lower and significantly higher driver reaction times were discussed. Finally Euro NCAP agreed on a reaction time of 1.2 seconds, which is exactly the value that ASSESS forwarded to them.”

2.4 WP 4 - Pre-crash evaluation
2.4.1 Introduction
Within WP4, four mayor activities were performed:
1)Definition of test scenarios and (draft) test protocol based on the accidentology provided by WP1
2)Update of the different test facilities to be able to conduct AEB testing according to the (draft) test protocol
3)Definition and development of a test target
4)Evaluation and refinement of the pre-crash test protocol, including an extensive R&R study

In the following, all activities and their main outcomes will be described.

2.4.2 Definition of test scenarios and (draft) test protocol
The overall WP4 test matrix has been established by considering only accidentology data. It is known that safety systems that are considered as state of the art at the start of the project, cannot manage the complete test matrix. Oncoming or intersection scenarios especially, but also cut-in scenarios may require technologies that are neither affordable nor widespread as of today. Therefore the test matrix shall be considered as a comprehensive overview of potential use cases, which are also applicable to future systems; for current generation systems, only a part of the test matrix can apply.

For the definition and specifications of test scenarios and related manoeuvres that were defined within this protocol, the following boundary conditions were considered:
-All test scenarios and their respective maneuvers were to be based as good as possible on the accidentology analysis performed by WP1, the results of finalized and ongoing EU and other (inter)national projects and the activities of other organizations such as NHTSA. Only information available at the time of the definition of the scenarios was considered.
-No single vehicle accidents (loss of control) or accidents with vulnerable road users involved were to be considered.
-The scenarios and their respective maneuvers were based on the most significant accidents / conflict situations as specified and ranked by WP1.
-The final test and assessment program needed to be cost efficient; meaning that it shall be performable within approximate two test days, excluding preparations and reporting. This resulted in a final test program of 56 tests.
-Testing shall take place under 'normal environment conditions'. Environment conditions shall be recorded per test.
-For the whole test program only one target vehicle was considered. Multiple targets could be considered for future work.
-The specified scenarios and their respective maneuvers shall be challenging for the systems under investigation and a rating system should be possible. This means that tests defined in the preliminary test program that can be passed by default by each system were to be removed from the final test program. No such test was found within the established test matrix. Hence, there was no need to remove any of the tests.
-The testing must be performed safe for test-driver, vehicle under test and the environment.

No statistic approach as used by NHTSA was chosen for the test program as this would have extended the program significantly. Hence all tests needed to be repeatable and reproducible to ensure a system will not pass or fail based on test specification variations. This evaluation was conducted within task 4.3 and the results were satisfactory.

This test protocol preceded the one established from Euro NCAP and was given to Euro NCAP for consideration. Though not the entire ASSESS test procedure was included into the upcoming Euro NCAP AEB protocols, parts (as the general set of manoeuvres or the driver reaction time) were taken on board. Where Euro NCAP in the end went for a procedure with increasing test speeds as main variable between the manoeuvres, ASSESS set up a protocol with less variation in test speed taking into account more variables with different challenges as driver reaction time or overlap.

2.4.3 Update of different test facilities
Based on the scenarios identified and the driver reaction model defined, IDIADA, BASt and TNO updated their facilities to run the specified rear-end pre-crash performance tests.

At BASt, a remote-controlled kart system able to carry the rear end part of the ASSESSOR is used for testing. Status as of December 2010 (Deliverable D4.2) was as follows:
-Remote control only operated manually
-No actual distance information was available to the kart operator
-No driver reaction was implemented in the vehicle under test
-Crashability of the test setup was available up to 30 km/h impact velocity

Major improvements have moved the kart and ASSESSOR system crashability to 40 km/h with minor damage on the vehicle-under-test. A speed controller on the kart is able to control the speed with an accuracy of 1 km/h. A display in the vehicle-under-test (VUT) displays the actual velocity, relative velocity and distance in x-direction to the kart operator. All these quantities are calculated from GPS position and speed readings. Braking control is done via an open-loop control, and lateral motion is still controlled completely by the kart operator (for safety reasons).

IDIADA used a rabbit vehicle during previous tests, as specified in D4.2 the rabbit mechanism was updated in order to reach higher impact speeds.

The previous system was a mechanical trigger released by the impact force. The new mechanism uses electromagnets to quickly release the target. The magnets are controlled by a microprocessor monitoring 2 parameters:
1.Acceleration of both rabbit and target vehicle, if a difference higher than a limit is found the controller releases the target.
2.Touch sensor signal for the target: If the rear end of the target is touched then the target is released.

2.4.4 Definition and development of a test target
One of the main items was the development of the test target which has to mimic the image of a passenger car as perceived by sensors used in common pre-crash systems. As the target is impacted by the vehicles under test (VuT) it should be soft to avoid damage to the VuT upon impacts up to 50 km/h or higher. Humanetics devised a full size target called ASSESSOR consisting of vented boxes for front, rear and side parts of the vehicle including appropriate RCS and optic features like lights and license plate.

2.4.5 Evaluation and refinement of the pre-crash test protocol including extensive R&R study
The test program was conducted with 4 cars in 3 different labs (IDIADA, TNO and BASt) with different initial approaches. Additional tests were conducted at Daimler and ADAC to include even more labs. Brake robots were installed in test vehicles to mimic repeatable driver behaviour (braking) under warning.

2.5 WP 5 - Crash evaluation
In order to understand the effect of the pre-crash systems activation on the occupants of the vehicle during the crash phase analysis were made of the occupant motion under pre-crash braking and possible means to restrain the occupant in this phase (pre-pretensioner). The following activities were performed:
-Braking manoeuvres: Braking tests were conducted with volunteers to analyse the forward motion under decelerations from braking. In addition numerical simulations using active human body models were performed.
-Crash simulations: Simulation of the crash phase were performed using detailed interior models. Impact speed reductions and out of positions due to the pre-brake actions were considered. The influence of the pre-pretensioner activation to keep the occupant in position was considered.
-Sled tests: Next to the simulations sled tests were performed focusing on the crash phase in a full frontal impact against rigid wall scenario. Means to represent the occupant forward displacement from pre-braking were introduced by adding foam between the belt and the dummy. Tests were done with and w/o pre-pretensioning. Speed reductions due to the braking action were taken into account. The sled tests were done using a buck of the same vehicle as used in the crash simulations listed above.
-Full Frontal impact test: One full frontal impact test was performed with full activation of the pre-crash braking and pre-pretensioner. The benefit coming from the pre-safe activation was analysed by comparing the results from this test with the ones from a reference test.
-Offset Deformable Barrier Impact Tests (ODB): Two ODB impact tests with pre-brake action were done, one activating the pre-pretensioner and the other one without activating it. By comparing these tests with the official Euro NCAP, it was be possible to analyse the effect of the two pre-crash systems (pre-brake and pre-tensioner) separately. For these tests a small class vehicle was used different from the vehicle used in the above activities.

2.5.1 Forward motion analysis: Braking manoeuvres and numerical simulations
In order to better understand the forward movement of the occupants of a car braking tests were performed with volunteers and numerical simulations with active (muscle tension) human body models.

2.5.2 Crash simulations
The objective this simulation study was to investigate the effect of reduced velocity on dummy injury values taking into account different dummy pre-crash positions. Results of crash tests and sled tests were available for similar configurations with different impact velocities. The main advantage of the simulations is that the dummy model can be positioned for pre-crash occupant motions. This is more complex in sled and full scale tests.

2.5.3 Sled tests
Whereas the numerical simulation analysis could provide first results for the effects of pre-braking with and without pre-pretensioning using dummy models and even human models, sled tests were performed to complement and extend those results. In order to be comparable to the simulations as well as to the final full frontal crash, the same vehicle base was used. Based on the experience from several pre-braking tests in real cars with humans and dummies, it was observed, that the forward displacement during braking is quite different between humans and Hybrid III dummies. Furthermore, on the driver side additional stabilizing support for the human/dummy can be observed, provided by the hands on the steering wheel and the feet on the pedals. Since this effect could not be studied in detail in this project but would significantly influence the results, it was decided to concentrate on the passenger side although tests were performed for the driver side as well.

2.5.4 Full Frontal impact test against rigid wall
Next to the simulations and sled testing full scale tests were performed to compare the performance with and without pre crash braking and pre/pretensioner. Data from a reference test without pre/crash system activation was already available. The second test was performed at BASt with the support of Daimler. The setup was the same as in the reference configuration, with a pre brake action and pre-pretensioner activated. For the braking the vehicle was equipped with a brake robot. The activation of the electric belt pretentioner was done by a sudden brake action of the brake robot.

2.5.5 Offset Deformable Barrier Impact Tests
In addition to the complete set of activities described in the previous sections on the large passenger car crash tests and simulations were performed for a small car as well in a Euro NCAP offset test configuration. As the reference test was available two other tests were done to study the effect of pre-crash braking and pre-pretensioners:
-One test with pre-brake action but without improved restraint systems activation.
-Another test with pre-brake action and improved restraint systems activation.

2.5.6 Summary of findings from crash analysis
In this part of the ASSESS project the main passive safety tools were used in order to understand the effect of the Integrated Safety System (ISS) activation during the crash phase. As far as the influence of such systems on the crash performance is concerned it was found that all tools clearly showed that the lower impact speed resulting from the pre-brake action reduced substantially the biomechanical values of the vehicle occupants. The activation of the pre-pre-tensioner seems to be also beneficial to reduce the injuries on the occupants of the vehicle, but in a lesser extent than the pre-brake action. An improvement of the dummies' readings would be possible by adapting the restraint system to the new energy level, which has been reduced by the pre-brake action.

2.6 WP 6 - Dissemination
An overview of the main dissemination activities is provided in section 3.2.

Potential Impact:

3.1 Potential Impact
Of all transport problems, safety is the one with the most serious impact on the daily lives of European citizens. In 2009, more than 35,000 people died on the roads of the European Union, i.e. the equivalent of a medium town, and no fewer than 1,500,000 persons were injured. The cost for society is huge, representing approximately 130 billion Euro in 2009. In view of this the European road safety policy orientations up to 2020 new technologies that have high potential to improve road safety should be promoted. This includes Integrated Safety Systems (ISS) like pre-crash systems with anti-collision warning or autonomous vehicle actions. For such systems it is stated that 'Accelerated deployment and broad market take-up of such safety enhancing applications needs to be supported in order for their full potential to be unleashed'. The ASSESS project is directly responding to this by developing test and assessment procedures for the evaluation of such systems

3.1.1 Societal impact of ASSESS - improving road safety
Integrated Vehicle Safety Systems (IVSS) - that combine elements from active and passive safety - have a high potential to improve both comfort and safety of vehicles and their occupants . To realise the full benefits, however, the new systems have to be widely deployed in the marketplace. That is a key element and its only when a significant number of vehicles on the road are fitted with these advanced technologies that a critical 'system effect' will be achieved. Unfortunately, there is still a large gap between technology development of IVSS and its deployment at a reasonable cost because of legal barriers (liability risk), the relatively high cost of such systems, the lack of information, throughout society, about the potential benefits of IVSS and the consequent lack of customer demand.

ASSESS addressed these topics by developing a relevant set of test and assessment methods applicable to integrated vehicle safety systems that are nowadays readily available on the market namely: Autonomous Braking Systems. The gained knowledge was implemented in test and assessment procedures which were forwarded to Euro NCAP as input to the definition of test protocols. This new Euro NCAP protocol for AEB testing will become effective in 2014. The ASSESS consortium is strongly convinced that the introduction of these protocols and the contributions from the ASSESS project to these protocols will further reduce fatalities and injuries. It is well known that Euro NCAP procedures have strongly influenced manufacturers to build vehicles that consistently achieve high ratings, thereby increasing the safety of vehicles .

By defining protocols and contributing to the implementation of the 2014 Euro NCAP protocols the following impacts as defined in the proposal are still regarded to be realistic:
-Acceleration of the introduction of integrated safety systems (ISS) and pre-crash sensing systems in particular that offer optimal protection in frontal collisions (in rear end, frontal and junction accident scenario's) into the EU vehicle fleet to a level of over 60% of all new vehicles in 2020 (meaning within two new vehicle generations time).
-To improve the survivability in frontal collisions by 20% and a reduction of the risk of severe injuries by 25% by 2020, reported and monitored through the European Road Safety Observatory

3.1.2 Environmental impact of ASSESS
Accidents are one of the main causes for traffic jams; between 12 and 15% of the (motorway) congestion in the Netherlands is caused by accidents (the 2nd cause after simple shortage of road capacity) . In a study by TNO on measuring emissions at highways it became clear that the emissions of both NOx (mainly NO2) and particles (PM10) could double in case of a traffic jam . Since ASSESS will have a positive effect on road traffic safety by the accelerated introduction of IVSS in new vehicles it will consequently contribute to the reduction of local harmful emissions like NOx and PM10, but also green house gas emissions (i.e. CO2). Furthermore, it is the strong belief of the participating OEM's that weight reduction of the vehicle's crash structure can be realised as a consequence of the introduction of IVSS in new vehicles. This will consequently result in lower fuel consumption and lower emission of CO2.

3.1.3 Strategic and economic impact of ASSESS
A direct measurable cost of road accidents is in the order of EUR 45 billion Euro. Indirect costs (including physical and psychological damage suffered by the victims and their families) are three to four times higher. The annual figure is put at EUR 160 billion, equivalent to 2% of the EU's GNP . The sums spent on improving road safety fail to reflect the severity of the situation outlined above. Efforts to prevent road accidents are still inadequate, corresponding to less than 5% of the total costs of those accidents, i.e. EUR 8 billion. Based on the realization of protocols and their implementation in NCAP by 2014 the outcome of the ASSESS project will significantly contribute to improving road traffic safety and herewith contribute to the reduction of the societal cost of road accidents.

3.2 Main dissemination activities
The following sections provide an overview of the main dissemination achievements during the project and plans after ASSESS is finished. Note that the project website is presented in Chapter 4 of this report.

3.2.1 Project public reports
Dissemination of projects results by making deliverables publicly available is regarded as one of the most important means to publish results. For that reason all technical deliverable reports are either marked as public or a public version for download from the website is generated.

3.2.2 Project flyers and newsletters
To bring the ASSESS project results under the attention of the automotive community flyers and newsletter were prepared during the runtime of the project. A first flyer was developed in M6 containing a brief overview of the project main goals, the technical approach, the expected achievements and a list of project participants. Next to this annual newsletters with main results and findings were prepared and distributed. Finally to promote some specific results flyers on dedicated topics were generated. This includes a flyer on the ASSESS test target with background information (specifications etc.) and status of the target. All flyers and newsletters were distributed via the dissemination database, the ASSESS website and via partners websites.

3.2.3 Supervisory Board meetings
During the first year of the ASSESS project a Supervisory Board consisting of high level representatives from Industry and Governments was established. The main role of this Board was to reflect on project findings and advise project members on directions for future activities.

3.2.4 Public workshops
A mid-term workshop was held on October first, 2010 at the IDIADA premisis in Spain. Over 20 representatives from related projects, industry and governments world-wide joined the workshop. ASSESS results were presented (morning session) as well as results from related projects (afternoon). Apart from ASSESS results from the projects vFSS, CAMP (US based project from NHTSA and OEM's) and eVALUE (EU FP7 ICT) were presented.

3.2.5 Participation and/or Organisation of Events
Harmonisation Platforms
Because of their potential Euro NCAP plans to introduce testing procedures for Autonomous Brake systems in 2014. Procedures are being defined and implemented using information from a number of projects:
-Advanced Forward-Looking Safety Systems (vFSS)
Cooperation between OEMs, research and insurance groups world-wide developing test and assessment methods for forward facing safety systems related to accidents with pedestrians and cars. vFSS also develops and applies methods on system effectiveness.
-Advanced Emergency Brake systems (AEB)
Cooperation between insurance organisations Thatcham and IIHS with support from research groups, a supplier and two OEMs. Aims and goals identical to vFSS.
-Assessment of Integrated Vehicle Safety Systems (ASSESS)
EU FP7 Project consortium of OEM's, suppliers, test houses, research organisations and universities. Total 14 partners. Research on test methods for car - car accidents (no pedestrians) considering driver behavioural aspects (warning), pre-crash performance evaluation, crash performance evaluation and system effectiveness.
-Allgemeiner Deutscher Automobil-Club (ADAC)

Cooperation with projects from other regions
In view of the technical status (close to broad market introduction) and the potential of Integrated vehicle Safety Systems a number of groups world-wide are developing test and assessment procedures. In the US the CAMP initiative from NHTSA and the US car manufacturers developing test procedures for the pre-crash evaluation. This so-called CAMP-CIB (Collision Imminent Baking) project finished end 2010 and forwarded results to ASSESS during the Mid-Term Workshop held September 2010. The CAMP-AEB project joined the second Supervisory Board meeting held in November 2012 at BASt.

Work Package 5 dealing with crash performance evaluation discussed with JAMA / JARI (Cooperation of Japanese OEM's and Research groups respectively) possible cooperation on this complicated topic. However, as decided during the second Supervisory Board meeting no specific testing related to passive safety was to be included in protocols. Instead benefits and risks related to pre-crash braking was to be exploited including recommendations for passive safety devices to be used when applying pre-crash braking.

ActiveTest is a Coordination and Support action running under DG ICT. This project is organising workshops and events related to testing of IVSS systems. ASSESS participated in two out of the three ActiveTest events organised so far. During the second ActiveTest event in September 2011 a project overview and detailed status per work package was provided by all WP leaders. In the third ActiveTest event held September 2012 special sessions were arranged as the ASSESS final workshop. Over 70 representatives from Industry, governments, research groups and academia world-wide joint this event.

3.2.6 Conference and Journal Papers
During the runtime of the project papers were presented at key conferences in the field of vehicle safety like ESV, ESAR/IRCOBI, FISITA and TRA conference.

3.2.7 Planned dissemination plan for the period after the project's runtime (2012-2016)
An overview of intended dissemination activities by project partners for 2013 - 2016 is provided below.

2013 - 2016 Present results of research activities as follow-up at international conferences, e.g. ESV 2013 and through the project's website:
Results on driver reactions, accident scenarios were presented at leading conferences in the field of vehicle safety including ESV. For the 2013 ESV abstracts were submitted and accepted to present results. In particular attention will be given to the WP5 activities on passive safety related aspects. This work was largely conducted during the last year of ASSESS and not presented in detail at any conference.

2013 - 2016 Possible implementation of the new test, evaluation and assessment methods of IVSS at Euro NCAP. Continued with possible implementation at other regions of the world consumer test organizations

Euro NCAP is defining protocols for Advanced Emergency Brake systems (AEB systems) for rear end impacts. The protocols will become effective in 2014. As reported above ASSESS partners BASt, IDIADA and TNO which are also representatives in Euro NCAP have been discussing the new protocols over the past years with Euro NCAP. Information and results from ASSESS have been forwarded to NCAP and many findings from the project will be implemented in the protocols. This includes for instance the driver reaction model.

The protocol definition is in its final stage with refinements from evaluation testing by all NCAP labs to come. After 2014 Euro NCAP will continue to develop the protocols for a phase II implementation by 2016 or later. This may include for instance other crash modes than rear end shunts. In this phase the ASSESS test target might become reintroduced.

2013 - 2016 Availability of the project website, to keep results and findings directly accessible to the public
To provide access to the results and findings the ASSESS website will remain available for the next four to five years.

2013 - 2016 Presentation of results to stakeholder groups like ISO and GRSP
Depending on future plans of ISO and GRSP related to IVSS systems like AEB results of ASSESS will be presented to these groups as input to future regulations. As off now no specific GRSP group related to AEB regulations has been formed. ISO is considering to form a group though. In case a GRSP group will be established it is very likely that research groups like BASt, IDIADA, TRL, TNO and Chalmers will present their findings to such a group via their National Governments. Also OEM's Daimler, Toyota and PSA as well as suppliers Bosch and TRW will forward results via ACEA and CLEPA respectively.

3.3 Exploitation of results
The future introduction of integrated vehicle safety systems in the market will very strongly depend on the availability of functioning and cost efficient products from the suppliers. OEMs are in need of reliable methods to evaluate these functions within their area of application to sell the functions to their customers, herewith creating demand for Integrated Vehicle Safety systems. The system suppliers will play an important role in accelerating market introduction by (re)designing their systems based on the ASSESS results. At the same time targets (standardisation and regulatory requirements) will be defined for the evaluation of such systems to further facilitate the introduction into vehicles. Project results on the assessment of integrated vehicle safety systems and the usage of assessment tools herein need to be transferred (discussed and agreed upon) to policy makers for implementation.

In view of the above main exploitable results from the ASSESS project relate to know-how and experience gained on defining and implementing the test procedures. This know-how was already and will in the future be used in consultation with policy makers when developing protocols and regulations (Test houses, OEM's, suppliers), consultation between OEM's and suppliers when specifying systems (OEM's, suppliers), consultancy work for governments and industry (test houses).

Some specific examples include:
-know-how on accident analysis and benefit studies: test houses advice to governments and use in consultancy work for industry; OEM and supplier internal use for function specification, use in communication to stakeholder groups
-Specifications of test scenarios and test conditions: OEM and supplier use for support to discussions with regulatory and consumerist bodies as well as specification and verification of the relevance of in-house test cases / scenarios; test houses advice to governments and use in consultancy work for industry.
-Practical know-how and experience from pre-crash and crash testing: test houses providing services on automotive design and testing, advice to / stakeholder groups governments as input to regulations and consumer protocols.
-Know-how on occupant pre-crash kinematics: test houses doing consultancy services on automotive design, safety system design.

List of Websites:

Related documents