Seminar

Relevance in Connected and Autonomous Driving
Luca Lusvarghi

Abstract: Connected and autonomous vehicles leverage Vehicle-to-Everything (V2X) communications to continuously exchange detected objects (e.g., surrounding vehicles or pedestrians) and cooperatively extend their field of view beyond their onboard sensors' line-of-sight range. While some objects are key to planning safe and effective maneuvers, others may not be relevant and have no impact on the vehicle’s driving decisions. The exchange of irrelevant objects has a two-fold impact: first, it unnecessarily consumes communication resources, potentially congesting existing V2X networks and hindering the large-scale deployment of connected and autonomous driving; second, it unnecessarily increases the computational strain on the vehicle’s hardware (CPUs, GPUs), ultimately increasing energy consumption, onboard processing latencies, and the noise injected into downstream planning tasks.

Understanding the objects’ relevance is thus key to optimizing the usage of communication and computational resources and enabling the design of more scalable and effective connected and autonomous driving systems.

To this end, this talk will present novel relevance estimation techniques able to capture the relevance that surrounding objects have for a connected and autonomous vehicle. Through a set of preliminary results, it will also shed light on the key contextual elements that can influence the relevance of an object and illustrate the potential benefits that a relevance-aware filtering of the exchanged objects can have in terms of computational and communication efficiency.

Speaker bio:

Luca Lusvarghi is a postdoctoral researcher at the Networked Systems Lab of the Universidad Miguel Hernandez de Elche (Elche, Spain). His research work lies at the intersection of wireless communications, intelligent transportation systems, and Artificial Intelligence (AI), with a particular focus on relevance estimation for connected and autonomous driving as well as semantic and task-oriented Vehicle-to-Everything (V2X) communications. His work is currently supported by a 2023 Marie Sk?odowska-Curie Actions (MSCA) postdoctoral fellowship. He received his master’s degree (summa cum laude) in electronics engineering and the Ph.D. degree in ICT from the University of Modena and Reggio Emilia (Modena, Italy) in 2019 and 2023, respectively. He was the recipient of the Vetrya Award for the best Italian Master’s degree thesis on 5G and of the GTTI PhD Award for the best Italian Ph.D. thesis in ICT.