Project

Measures for behaving safely in traffic MeBeSafe

Goal: Of all transport modalities, road traffic is by far the most dangerous . In 2014 almost 26,000 people were killed and 300,000 seriously injured on EU roads . The major cause in most road accidents is “user error”—which we define as traffic user behaviour inappropriate to the risk posed by the situation, reducing safety margins to zero. Almost all serious road accidents involve at least one driver --serious accidents between Vulnerable Road Users (VRUs) are rare. Safer drivers will have a positive impact on all road users, reason why this project’s primary focus is on changing driver behaviour.
A large number of measures have already been proposed and/or implemented that intend to reduce the occurrence of these well-known user errors or mitigate against their consequences. A range of in-vehicle measures (Advanced Driver Assistance Systems—ADAS) assist ad-hoc driver decision-making or in some cases act autonomously under driver supervision. Prior studies show the potential of these measures, but also show that drivers are not using them to their full benefit—they are frequently switched off or their feedback is ignored.
For most of us, navigating traffic is a very common activity—so common that a large part of it is habitual. Almost automatic. We seek to change this habitual behaviour (rather than just the emergency behaviour targeted by autonomous ADAS) in order to increase safety margins. Our proposal focusses on a category of “nudging ” feedback measures that aim to guide the user towards desired behaviour yet preserve his freedom of action—mainly targeting car drivers but also exploring the potential to direct the behaviour of cyclists.
Autonomous safety measures deploy in critical situations and need to perform the right action 100% of the time, to gain widespread driver acceptance—especially in the light of the expected move towards more autonomous vehicles in the traffic mix. Feedback (keeping the driver in the loop) is given earlier and more often, before the situation gets critical. Feedback is both easier to implement (a lower confidence level is required compared to autonomous vehicle action), and is likely to find higher acceptance (as it keeps the driver in control). For all but the highest level of automated driving, driver feedback will remain an important aspect of safe behaviour in traffic.
Road accident statistics clearly show a number of high-level causation factors: Failure to look properly (lack of attention); Excessive speed for the circumstances (leading to loss of control and failure to timely spot hazards); Impeded mental and/or physical condition of driver (drink driving, fatigue); It is these already identified high-level risk factors that MeBeSafe intends to address.
Many suggest that the best way to reduce human error in traffic is to remove the human driver and move to a fully autonomous fleet. At best this will be a long time in the future, much longer than the timeframe the measures suggested in our project will make their impact felt. Instead of absolving the driver from responsibility, MeBeSafe seeks to provide objective, behaviour-changing feedback to steer drivers towards preserving adequate safety margins. Over the longer term this changes habits and makes the user’s behaviour intrinsically safer.

Date: 1 May 2017 - 1 December 2020

Updates
0 new
7
Recommendations
0 new
1
Followers
0 new
23
Reads
0 new
559

Project log

Anna-Lena Köhler
added a research item
Daily driving as a relatively well-trained process is based on both, conscious and sub-conscious decision-making processes and auto-mated/reflective behaviours. These habituated behavioural patterns and habituation effects can lead to uninten-tional traffic violations due to late intervention. In these cases today's conventional traffic interventions are insufficient. The H2020 project MeBeSafe develops and tests innovative measures to increase safety margins according to the concept of nudging. The central approach is to stimulate the driver to show a desired choice without forbidding alternatives. As part of the project, dynamic lighting systems were installed in Eindhoven's infrastructure and examined in field tests for their suitability as speed reduction measures. Additionally, an in-vehicle nudging measure has been developed to alert the driver to potentially dangerous situations through an in-car human-machine-interface. In this paper, the application of such implementa-tions in infrastructure and vehicle development will be discussed on the one hand and potentials and obstacles associated with the implementation on a larger scale will be examined on the other hand.
Stefan Ladwig
added 2 research items
After over 130 years of motorized road transportation, daily driving can be defined as a relatively well-trained process, relying on habitual patterns. These are based on a sound proportion of both, elaborate (sub-) conscious decision-making processes and automated/reflexive behaviors. However, apart from human factors, navigating safely through daily traffic inevitably needs additional guidance by road traffic regulations. On a macroscopic level, the driver´s acquaintance with such rules is precondition and needs to be verified by a valid driver´s license. However, accident statistics show that even in times of supportive ADAS and various traffic interventions like rumble strips, radar, intelligent traffic signs etc., a high number of severe road accidents appears to be still an issue on European roads. The authors follow the approach that conventional traffic interventions are insufficient when unintentional violations are concerned and hence, appear to be ineffective due to the following reasons: First, they do intervene too late in time. For instance, getting a fine due to speeding on a certain road section happens late after the misbehavior has occurred. Second, conventional traffic interventions lack effectivity due to habituation effects. In this vein, the present paper empirically surveys innovative measures following the concept of nudging in order to increase safety margins. This concept has been adopted from behavioral economics and relates to subconsciously stimulating humans to make a desired choice without prohibiting alternative choices. Nudging works on a subliminal level, is less invasive, provides humans with a choice (by predisposing a desired choice) and can be applied both, earlier as other measures as well as on demand in the chain of events leading to critical situations. The present paper targets driving at an inappropriate speed as of several causes for dangerous situations and accidents. Following the concept of nudging, the authors assume that dynamic infrastructure nudging solutions explicitly targeting drivers going at an inappropriate speed, have the potential to make these drivers adjust their speed to an appropriate level in a more effective way than conventional road signs do. Hence, the authors hypothesize that light-emitting nudging solutions have a stronger impact on drivers to reduce speed in a given situation than non-dynamic measures do. Second, they hypothesize that dynamic nudging solutions using moving lights towards the driver have a stronger impact to reduce speed in a given situation than non-moving lights do, because of the visual impression of the driving speed to be higher than it actually is. The empirical experiment consisted of three conditions. Condition 1 was a baseline showing conventional road signs indicating the desired speed level without any nudging measures. Condition 2 and 3 consisted of light emitting spots mounted onto the road. The spots were programmed either to be statically illuminated (condition 2) or to move dynamically towards the drivers in order to slow them down by creating a visual illusion of driving faster than they actually do (condition 3). In a driving simulator study, N = 54 participants drove through a simulated real-life motorway exit where dangerous driving behavior had been spotted before. The results show that participants reduce their speed significantly in the nudging-conditions in comparison to conventional measures. Furthermore, the hypothesis stating that moving lights toward the driver have a higher impact on speed reduction than non moving-lights can be confirmed as well. Further specification and testing is needed to determine the final design of such measures and their generalizability to other traffic situations. Furthermore, the influence of such new stimuli on the choice of an appropriate trajectory depending on the speed in a hazardous situation will have to be investigated.
Daily driving can be defined as a relatively well-trained process, relying on habitual patterns. These are based on both, conscious and subconscious decision-making processes and automated/reflective behaviours. Today's conventional traffic interventions are insufficient when unintentional violations are concerned due to a too-late intervention or habituation effects. The H2020-project MeBeSafe provides innovative measures to increase safety margins following the concept of nudging, which relates to stimulating humans to make a desired choice without prohibiting alternatives. Measures to reduce speeding through nudging via a dynamic light system in the infrastructure have now been installed for field testing in Eindhoven. Additionally, an in-vehicle nudging measure has been developed to direct the attention of the driver to potentially hazardous situations through an in-car human-machine-interface. This paper evaluates the application of these measures for future infrastructure and vehicle development and discusses the opportunities and inhibitions that come along with implementation on a larger scale.
Stefan Ladwig
added an update
This report describes different ideas for nudging solutions that can be implemented in vehicles to nudge the driver to:
o Make better use of safety functions onboard state-of-the-art vehicles that are equipped with various advanced driver assistance systems. The ideas for making better use of safety functions will be elabo@rated as part of the MeBeSafe coaching framework in WP4. In this report only an introduction to this type of in-vehicle solutions has been given.
o Direct their attention to potential hazards on the road. As a use case for this type of nudging, we focus at the interaction between cyclists and passenger cars on the road; representing a large number of traffic casualties which is difficult to address by current advanced driver assistance systems due to the high maneuvrability of cyclists.
To direct the attention to potential hazards, two basic system components are needed: 1. A model to estimate the level and type of hazard and 2. A human-machine-interface to provide appropriate information regarding this hazard to a driver. The report describes the set up of such a hazard prediction model and its components: a static world model referring to road layout and traffic rules, a dynamic world model that considers the actual detections of potential hazards on the road, and a cyclist trajectory prediction model that is intended to predict where a cyclist is going in the upcoming couple of seconds. Al the information from these components is integrated to estimate a hazard level in an approach of a cyclist intersection.
Moreover, different options for transferring information regarding the estimated hazard to the driver have been identified. The report shows which design rules and approach need to be followed to develop these options into an effective human-machine-interface.
Both in the hazard model as in the HMI-options, there is room for selecting parameter values that influence the effectiveness of the combined nudging-solution. In a next step in MeBeSafe WP2, tests with simulations and test with simulators will be used to determine the most promising in-vehicle nudging solution that will actually be implemented as a prototype in one FIAT 500X vehicle for testing in WP5.
 
Stefan Ladwig
added an update
This document describes the research methodologies employed within Work Package 4 "Driver Coaching". This work package focuses on the development of driver coaching schemes, supporting coaching software/apps, and evaluations of such systems. Some of the inherent tasks are directly related to each other (e.g. concerning coaching of Heavy Goods Vehicle drivers), such that their research methodologies are closely aligned; whereas others can and will be executed more independently (e.g. concerning coaching private vehicle drivers on the use of ACC) and therefore have their own methodology. For all tasks, we distinguish between the methodology for development and the methodology for evaluation, each of which is described in some detail.
Each of the research methodologies has at this point been sufficiently defined and where necessary aligned, such that we can move forward with the development and small-scale evaluation of the coaching methods and apps within WP4, and subsequent larger-scale evaluation in the field test of WP5.
 
Stefan Ladwig
added an update
One of the objectives in MeBeSafe is the coaching of heavy goods vehicle drivers (HGV) on their own behaviour. It is well known, that risky driving behaviour is highly correlated to severe chrashes - however, by utilizing a tailored coaching methodology risky driving behaviour will be prevented and will therefore contribute to safer traffic behaviour.
The present deliverable serves as a progress report. Objective was the investigation on what data will be needed for coaching purposes, how it could be gathered, what variables could be rated to be relevant and how these can finally be used for driver profiling.
With regards to technology, our recommendation is to collect data on driving behaviour and driving context by means of a mobilephone, augmented with inward- and outward-facing cameras. In terms of driver profiling we aimed to capture "the tendency to behave a certain way in a certain situation of context" and distinguish meaningfully between diverse contexts/situations in which a particular type of behaviour occurs. Therefore driver profiles were developed using driving behaviour variables measured by telematics, including context information. "The Traffic Safety Wheel" was developed, representation of driver profiles where we can compare driver behaviour with fleet behaviour across varying driving contexts. Based on the results further decisions can be made on how to proceed in the MeBeSafe project.
 
Stefan Ladwig
added an update
Please visit www.mebesafe.eu and track our recent activities.
Leader of Deliverable:
MariAnne Karlsson, Chalmers University / SAFER
Contributors:
Rino Browers, TNO
Felix Fahrenkrog, BMW
Anita Fiorentino, FCA Italy
Tineke Hof, TNO
Anna-Lena Köhler, ika
MariAnne Karlsson, Chalmers/Safer
Mikael Ljung Aust, VCC
Anneli Selvefors, Chalmers/Safer
Ingrid van Schagen, SWOV
Angela Sprajcer, SHELL
Antonella Toffetti, FCA Italy
Divery Twisk, SWOV
Anders af Wahlgren, Cranfield University
Johann Ziegler, VUFO
Reviewer:
Niccolo Baldanzini, Univeristy of Firenze
Formal review and responsible for upload to EU:
Coordinator of MeBeSafe, ika
Maximilian Schwalm, Stefan Ladwig, Anna-Lena Köhler, Kathrin Hülsen
Integrated framework can be found here as well:
 
Stefan Ladwig
added an update
Dr. Phil. Mikael Ljung Aust and Prof. Dr. Phil. Maximilian Schwalm recently published an article on Nudging at 26th Aachen Colloquium Automobile and Engine Technology 2017. Please contact the authors for more details.
Abstract:
MeBeSafe is a H2020 RIA project funded by the European Commission. Its objective is to stimulate traffic participants toward improved safety margin preservation in common traffic situations which carry an elevated risk, i.e. improving their risk management. Most current approaches try to make people behave safer by explicitly appealing to reason. MeBeSafe on the other hand starts from the realization that traffic behavior is largely automated and habitual, with reason playing a limited role. The project therefore aims to change road user behavior by an alternative approach called nudging which stems from behavioural economics. It relates to (subconsciously) pushing humans in a desired behavioural direction through clever feedback and persuasive design, yet without being prohibitive against alternative choices of action.
 
Stefan Ladwig
added an update
Project Logo
 
Stefan Ladwig
added an update
Press Release on Project Kick-Off / End of June 17
 
Stefan Ladwig
added a project goal
Of all transport modalities, road traffic is by far the most dangerous . In 2014 almost 26,000 people were killed and 300,000 seriously injured on EU roads . The major cause in most road accidents is “user error”—which we define as traffic user behaviour inappropriate to the risk posed by the situation, reducing safety margins to zero. Almost all serious road accidents involve at least one driver --serious accidents between Vulnerable Road Users (VRUs) are rare. Safer drivers will have a positive impact on all road users, reason why this project’s primary focus is on changing driver behaviour.
A large number of measures have already been proposed and/or implemented that intend to reduce the occurrence of these well-known user errors or mitigate against their consequences. A range of in-vehicle measures (Advanced Driver Assistance Systems—ADAS) assist ad-hoc driver decision-making or in some cases act autonomously under driver supervision. Prior studies show the potential of these measures, but also show that drivers are not using them to their full benefit—they are frequently switched off or their feedback is ignored.
For most of us, navigating traffic is a very common activity—so common that a large part of it is habitual. Almost automatic. We seek to change this habitual behaviour (rather than just the emergency behaviour targeted by autonomous ADAS) in order to increase safety margins. Our proposal focusses on a category of “nudging ” feedback measures that aim to guide the user towards desired behaviour yet preserve his freedom of action—mainly targeting car drivers but also exploring the potential to direct the behaviour of cyclists.
Autonomous safety measures deploy in critical situations and need to perform the right action 100% of the time, to gain widespread driver acceptance—especially in the light of the expected move towards more autonomous vehicles in the traffic mix. Feedback (keeping the driver in the loop) is given earlier and more often, before the situation gets critical. Feedback is both easier to implement (a lower confidence level is required compared to autonomous vehicle action), and is likely to find higher acceptance (as it keeps the driver in control). For all but the highest level of automated driving, driver feedback will remain an important aspect of safe behaviour in traffic.
Road accident statistics clearly show a number of high-level causation factors: Failure to look properly (lack of attention); Excessive speed for the circumstances (leading to loss of control and failure to timely spot hazards); Impeded mental and/or physical condition of driver (drink driving, fatigue); It is these already identified high-level risk factors that MeBeSafe intends to address.
Many suggest that the best way to reduce human error in traffic is to remove the human driver and move to a fully autonomous fleet. At best this will be a long time in the future, much longer than the timeframe the measures suggested in our project will make their impact felt. Instead of absolving the driver from responsibility, MeBeSafe seeks to provide objective, behaviour-changing feedback to steer drivers towards preserving adequate safety margins. Over the longer term this changes habits and makes the user’s behaviour intrinsically safer.