#	Pagina
attuale pagina	/open-h2020/projects/210138/results.html

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - MeBeSafe (Measures for behaving safely in traffic)

Teaser

Summary

MeBeSafe focuses on behavioural feedback measures to be provided to vehicle drivers and cyclists, with the objective of stimulating safer behaviour in common traffic situations carrying an elevated risk, making users better preserve safety margins. Nudging measures will allow for choosing freely between different behaviours, but the choice is presented in a way to predispose users towards making a desired choice in the immediate situation. Coaching interventions aim for educating people towards adopting safer behaviour when or after a certain hazardous situation has occurred.
The overall objectives of the project include getting drivers to take a break, making them use ADAS to prevent close following, making them more attentive to potential hazards, achieve behavioural change via car- and HGV-driver coaching, prompting drivers and cyclists to reduce their speed in hazardous road sections, and guiding them along a safe trajectory.

Work performed

1 Integrated framework
T1.1 Types of interventions to be developed and implemented include nudging and coaching as well as a combination of which were typified in task 1.1. An integrated model is proposed. In addition to a literature review, further literature has contributed to identifying underlying theories and models of relevance for further understanding road user behaviour.
T1.2 This task included identifying and taking road user profiles or characteristics of relevance into consideration. In designing the user studies and field trials, demographical factors need to be controlled. There is limited knowledge on influence of profiles, but there are implications on cultural differences regarding openness to nudging.
T1.3 The design of interventions requires a process consisting of iterative steps and decision points. T1.3 involves formulation of design considerations, deduced from literature reviews, interviews, and workshops. The work has also resulted in some generic design guidelines.
2 In-vehicle nudging solutions
T2.1 The wireless information and communication equipment (WiCE) was further developed. Particular signal databases that include ACC state information and drowsiness monitoring were set up and made available to the prototypers. A sensor was integrated to determine the direction of driver attention in driving simulation. Simulator- and vehicle-tests proofed functioning.
T2.2 An architecture for model for cyclistâ€™s intent prediction was developed. Two observation studies were conducted. AI/machine learning techniques are used to determine typical cyclist manoeuvres. The model aims to predict manoeuvres relying on data from observing the trajectories of the cyclist over the last seconds in view of the vehicle. Interaction of cyclists with other road users is fed into the intent prediction. Relevant scenarios were selected. An HMI was defined. UDRIVE is used to select interaction scenarios for validation.
T2.3 The basic architecture for the world model was implemented. A hazard prediction model was constructed. The nudging system is backed up by an existing Cyclist-AEB system in a test car. A simulated environment for development support and to determine the difference between actual and perceived hazard was set up. Accident scenarios from the German GIDAS database are used to derive expectations for in-vehicle nudging solutions.
T2.4 An ACC awareness- and a drowsiness awareness HMI following D1.1 were developed. Potential nudging implementations of encouragement towards higher ACC usage, as well as making drivers take a break when drowsy were developed. For ACC nudging, 2 different nudging concepts were developed for trial. For drowsy driver nudging, a suitable in-vehicle interface was designed. A driving simulator test is currently prepared. Development is supported by data derived from UDRIVE. The HMI solution for directing driver attention is currently developed.
T2.5 Solution selection uses the CATS accident investigations, added by results from observation studies. Iterative development clinics are conducted. The HMI is prepared for installation in a driving simulator. A virtual testing environment to evaluate performance of nudging measures is set up. It will be used to run simulations to tune and calibrate system parameters that appear in hazard prediction model and HMI activation. The WICE equipment is currently installed in 5 pilot vehicles.
3 Infrastructure measures
T3.1 Experts designed first nudging measures. A first driving simulator study was completed. Data analysis identified promising nudging interventions. Light-emitting spots that fit the needs of the intended nudging measures are developed. Traffic flow in Eindhoven has been measured. Results ensure nudging measures meet real world requirements, backed up by statistics and contextual analysis as well as Monte Carlo-simulations.
T3.2 Workshops were conducted to generate ideas for speed reducing nudges. Experimental tests were done (indoor

Final results

The novelty of the current research is in the use of sensors and sensor data by algorithms that intend to predict the likelihood a detected and pre-defined situation leading to a dangerous one. Preserving safety margins reduces the risk (for the individual) and occurrence/severity of accidents (for society). MeBeSafe has brought the development in each WP to a level beyond the state of the art aiming at demonstrating their applicability in the field trials. Estimations re the socio-economic impact and societal implications are foreseen to be at hand after the field trials at the end of the project.