Skip to main content

Building Acceptance and Trust in Autonomous Mobility

Periodic Reporting for period 2 - Trustonomy (Building Acceptance and Trust in Autonomous Mobility)

Reporting period: 2020-11-01 to 2022-07-31

Despite technological breakthroughs in connected and automated transport, the transformation of existing transportation into a fully autonomous system will span several years. In the meantime, mixed traffic environments with semi-autonomous vehicles proactively passing the dynamic driving task back to the human driver, whenever system limits are approached, is expected to become the norm. Such a Request to Intervene (RtI) can only be successful and trusted by end-users if the driver state is continuously monitored and his/her availability properly evaluated and sufficiently triggered (through tailored human-machine interfaces - HMIs). In parallel, driver training has to evolve to account for the safe and sensible usage of semi-automated driving, whereas driver intervention performance has to be made an integral part of both driver and technology assessment. Besides, the ethical implications of automated decision-making need to be properly assessed, giving rise to novel risk and liability analysis models. The vision of Trustonomy (a neologism from the combination of trust + autonomy) is to maximise the safety, trust and acceptance of automated vehicles by helping to address the aforementioned technical and non-technical challenges through a well-integrated and inter-disciplinary approach, bringing domain experts and ordinary citizens to work closely together. Trustonomy investigated, setup, tested and comparatively assessed, in terms of performance, ethics, acceptability and trust, different relevant technologies and approaches, including driver state monitoring systems, HMI designs, risk models, and driver training methods. This has been performed through both simulator and field based studies, in a variety of autonomous driving and RtI scenarios, covering different types of users (in terms of age, gender, driving experience, etc.), road transport modes (private cars, trucks, buses), levels of automation (L3 - L5) and driving conditions.
The main results accomplished include:
• Achievement of a common vision for the successful investigation of Trustonomy research and innovation specific objectives, through the operational context definition, the elicitation of system and user requirements (functional and non-functional), the Trustonomy frameworks definition and a set of methodological guidelines.
• Definition of the specifications and the detailed and executive design of the Trustonomy frameworks.
• Implementation and delivery of the Trustonomy framework prototypes, developed in two consecutive iterations and delivered in three releases to be deployed and validated in the preliminary trials.
• Driver State Monitoring (DSM) Assessment Framework to evaluate the performance of DSMs functionalities by means of providing qualitative and quantitative reports.
• Assessment framework for Human-Machine Interfaces (HMIs), through an ergonomics software tool and a time-based approach
• An Automated Decision-Support framework for risk assessment, early warning, and trajectory planning
• Innovative approach to driver training with the development of a competence map, novel driver training curricula and an online e-learning platform
• Driver Intervention Performance Assessment (DIPA) framework to assess the take-over phase, comprising time-based intervention performance measures, as well as objective and subjective quality measures
• Trust and acceptance framework to investigate the main factors which can affect driver trust while interacting with an AV, to study its variations and how it can be influenced and reinforced
• Design, preparation, execution and evaluation of trials in 5 countries and different road vehicle types (driving simulators, cars, trucks and busses)
• Set up and consultation with the Trustonomy Advisory Board
• Successful project dissemination through targeted communication, webinars for Trustonomy Panel members and organization and participation at European mobility related events
• Collaborative design of a tailor-made exploitation strategy according to the project course of research, its outcomes and results of the trials
• Technological trends monitoring via the Innovation Management Observatory, with participation to Standardization working groups and clustering and relationships with other Cooperative, Connected and Automated Mobility projects
The main impacts from Trustonomy are:
• a methodological and technological framework for the operational assessment of different Driver State Monitoring (DSM) systems, enabling studies for the suitability of different DSM technologies.
• a framework for the operational assessment of various HMI designs, including visual, auditory, haptic, timing and content factors,to study their effectiveness in terms of ergonomics and influence on driver performance.
• an automated-decision-support framework covering risk assessment (precursors and forecasting models), to issue early Request to Intervene (RtI) warnings, and emergency trajectory generation to react to critical risks and RtI failures.
• novel driver training curricula for human drivers of ADS, according to emerging training needs in the non-fully autonomous era, based on risk mapping of CAD operation activities, addressing safe driver-vehicle interaction and making use of innovative, ICT-based driver training methods.
• a driver intervention performance assessment (DIPA) framework consisting in a set of widely acceptable methods comprising time-based intervention performance measures, as well as objective and subjective take-over quality measures.
• measurement of performance, trust and acceptance of human drivers of ADS, specifically investigating the mediating effects of the multi-dimensional aspects of trust on driver acceptance and engagement with autonomous vehicles, to assess innovative and safe ways of building, maintaining and regaining driver trust, focusing on driver intervention and transition of control scenarios.