Event

This event, organised by SAFER’s Working Group on Road User Behaviour, brings together partners to explore how emerging AI technologies – including conversational agents, large language models, and social robotics – can support safer road user behaviour.

Across the day, participants will gain insights from ongoing research and real-world applications that address key safety challenges, such as driver understanding of ADAS, impaired driver states like stress and drowsiness, and interaction between automated systems and human road users. The program combines technical perspectives with human factors, highlighting both opportunities and limitations of these rapidly evolving technologies.

A central part of the event is the interactive session, where participants will jointly identify relevant use cases, methodologies and key challenges. This creates a space not only to learn, but to actively shape future directions within the SAFER platform.

By participating, you will:

• gain a clear overview of how AI can support safer driver behaviour and interaction.
• understand current challenges related to trust, communication and system acceptance.
• explore how new technologies can be applied in real-world traffic contexts.
• connect with partners working on related topics and identify opportunities for collaboration.

The event is open to all SAFER partners and welcomes anyone interested in the intersection between human behaviour, AI, and traffic safety.

Register here before May 7 (SAFER partners only)!

TENTATIVE AGENDA

0900-0910: Introduction by Robert Lowe, Working group leader for Road User Behaviour at SAFER

0910-0950: "ConAI: AI-based conversational agents that support drivers' understanding of ADAS and enhance traffic safety" Johanna Vännström, Scania. 

The project ConAI aims to improve drivers’ understanding and safe use of advanced driver assistance systems through AI-based conversational support. In this presentation, we will highlight our approach to developing, prototyping and testing new concepts, along with the methods used to explore and evaluate their potential impact on truck drivers. The project is funded by FFI, coordinated by Traton and done in collaboration with RISE.

0950-1030: "I-AIMS (1-2): Impairment-Aware Intelligent Mobility Systems: Embodied LLM assistive agents for mitigating safety-relevant impaired driver states.", Robert Lowe, RISE/University of Gothenburg (or Henrik Lind, Smart Eye, tbd)

I-AIMS 1 and 2 assesses different scenarios in which LLMs with or without embodied interfaces can be used to mitigate safety-relevant driver states such as stress, drowsiness and negative affect. Such mitigation entails LLMs providing information about external events: i) anticipated safety relevant events (to anticipatorily mitigate negative affect), ii) imminient safety-relevant events (potentially imperceptible to the driver). It also entails providing information about driver states: i) in densely populated urban environments, e.g. related to stress and frustration, ii) in low-density populated environments, e.g. related to drowsiness. The effects of forms of embodiment (e.g. robot head) on road user safety are also evaluated. The project is funded by Vinnova, coordinated by the University of Gothenburg in collaboration with Smart Eye AB. 

1030-1045: Fika and networking opportunities

1045-1140: Interactive session: "Identifying use cases, methodologies and key challenges for using LLM and robotic agents to benefit road user safety" 

1140-1240: Networking lunch

1240-1320: "Enhancing Vehicle-Driver Interaction with AI for Better ADAS Acceptance?", Sylvie Wacquant, Magna

This presentation explores how AI can elevate vehicle–driver interaction to improve user acceptance of ADAS. It highlights persistent challenges, such as limited trust, unclear system intentions, and communication gaps, and examines AI-driven approaches that address these issues. These include personalized and context-aware assistance, natural language and multimodal interfaces, advanced driver monitoring, and coordinated use of interior and exterior sensing. By making interactions more intuitive, adaptive, and transparent, AI can strengthen driver trust, enhance safety, and accelerate the adoption of automated driving features.

1320-1400: "ROBOPOL: Social Robotics Meets Vehicular Communications for Cooperative Automated Driving", Alexey Vinel, Halmstad University (online presentation)

On the path toward full autonomy, heterogeneous traffic - where vehicles with different levels of automation and cooperation share the road with human users - remains unavoidable. Even in fully automated scenarios, vulnerable road users such as pedestrians will continue to shape traffic dynamics. In this context, we propose social robots as mediators between autonomous vehicles and humans, enabling intuitive, human-centered communication and influencing road user behavior in safety-critical situations. We first discuss key design enablers for such systems within a broader futuristic context of cooperative intersection management. We then narrow the focus to present-day traffic and introduce several proof-of-concept implementations, including pedestrian crossing assistance and support for vehicles exiting low-visibility areas. Finally, we focus on children’s safety and present results from two real-world experiments conducted in Europe: in Slovakia last year, where a humanoid robot interacted with school children in an outdoor traffic playground, and in Belgium this year, where two cooperative robots - a humanoid and a robot dog - supported children during pedestrian crossings near a school. These findings highlight the potential of social robotics to bridge the gap between automated systems and human road users and outline their emerging role in future traffic interactions.