Page 802 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

reduction of traffic-related deaths and injuries through intelligent systems that anticipate risks,
improve driver behavior, and enable faster emergency response.

AI-powered tools improve the accuracy of crash data collection and analysis, offering insights
that help prevent accidents and guide better infrastructure planning. These systems can support
dynamic traffic management and enable predictive interventions that reduce congestion and
increase overall transport system efficiency.

In the realm of intelligent transport, autonomous and semi-autonomous features such as lane-
keeping assistance, emergency braking, and automatic parking are already making vehicles
safer and easier to operate. These technologies also reduce driver workload and error, key
factors in many road incidents.

Moreover, intelligent transport benefits from AI-enhanced traffic coordination, enabling real-
time adjustments based on road conditions, vehicle density, and weather. This results in more
responsive and resilient urban mobility networks.

Emerging mobility solutions—such as connected vehicles, ride-sharing platforms, and integrated
infotainment and communication systems—further contribute to smarter transportation
ecosystems. These innovations help streamline vehicle coordination, improve route efficiency,
and create safer, more enjoyable travel experiences.
Ultimately, the adoption of AI in transport is not only about improving individual vehicle
performance but about building a connected, adaptive, and sustainable transport infrastructure
that enhances safety, accessibility, and urban quality of life on a system-wide level.

2�3� Future work

1. Data collection: The current dataset lacks quality and quantity. Hence, in the future, for
improved performance analysis, we need to have a dataset with a larger number of vehicles
in different platooning scenarios.

2. Proof of concept development
• Solving AI/ML Domain Adaptation Through Transfer Learning – Domain adaptation
is a key challenge in AI/ML, especially when models trained in one domain need to
generalize effectively to a different but related domain. Transfer learning provides a
powerful approach to address this issue by leveraging knowledge from a source domain
to improve performance in a target domain with limited labelled data.
• Making AI-driven systems more cost-effective and scalable – Explainable AI (XAI) can help
reduce computational costs in machine learning applications, including QoS prediction
and vehicular communications, by enhancing model efficiency, interpretability, and
optimization. With these advancements, the research can transition from theoretical
modelling to practical, real-world applications, enhancing the security of V2X systems.

3 Use Case Requirements

• REQ-01: It is critical to enhance Quality of Service (QoS) by defining the minimum
communication performance indicators, such as sub-10 millisecond latency and 99.999%
reliability[6], that 5G/6G networks and Multi-access Edge Computing (MEC) must meet
to support real-time decision-making in vehicular environments.
• REQ-02: It is mandatory to clearly specify the type of cellular connectivity required,
distinguishing between technologies like NR-V2X PC5 sidelink and Uu uplink, and to

766