Elishai Ezra Tsur ^1* Gianluca Di Flumeri ² Hadar Cohen Duwek ¹

• ¹ Open University of Israel, Ra'anana, Israel
• ² Sapienza University of Rome, Rome, Lazio, Italy

Autonomous driving is one of the hallmarks of arIficial intelligence. Neuromorphic, brain-inspired compuIng architectures, are revoluIonizing vehicular autonomy through biomimeIc approaches [1].They can dramaIcally impact the mulIdimensional landscape of autonomous driving, extending across criIcal domains such as control [2], navigaIon [3], and neurophysiological assessment [4]. ParIcularly, these brain-inspired computaIonal systems leverage spiking neural networks (SNNs) and event-driven processing to enhance real-Ime percepIon [5], decision-making [6], and adapIve learning capabiliIes [7].This arIcle collecIon revolves around the state of the art of neuromorphic systems, designed to support autonomous driving across those dimensions.Neuromorphic control is posed to significantly contribute to autonomous behavior by leveraging spiking neural networks-based energy-efficient computaIonal frameworks. In this collecIon, Halaly and colleagues explored neuromorphic implementaIons of four prominent controllers for autonomous driving: pure-pursuit, Stanley, PID (ProporIonal -Integral -DerivaIve), and MPC (Model PredicIve Control), using CARLA [8], a physics-aware simulaIon framework. Their results show that those neuromorphic control models converge to their opImal performances with merely 100-1,000 neurons.They also highlight the importance of hybrid convenIonal and neuromorphic designs, as was suggested here with the MPC controller. This MPC was later extended to support adapIve behavior and to address unforeseen situaIons such as malfuncIoning and swi` steering scenarios [9]. Further in this arIcle collecIon, Lian and colleagues addressed the importance of adapIve control for changing road condiIons by proposing a neural model for deriving road adhesion coefficient (the maximum fricIon coefficient between Ire and road surface) and Ire cornering sIffness (affected by road fricIon and Ire slip angle).Those parameters can be used to improve the design of the vehicle's dynamic model and enhance the controller's robustness.Autonomous driving systems (ADSs) o`en comprise mulImodal sensing, including cameras, LiDARs, IMUs, and GPS, for real-Ime object detecIon, semanIc segmentaIon, planning, and control [10]. LiDAR-driven neural percepIon is an important stepping stone toward the design of self-navigaIng vehicles [6]. In this arIcle collecIon, Lee and colleagues propose a neural reinforcement model, they termed the "Velocity Range-based EvaluaIon Method" in which LiDAR data is used to provide path planning in a map-less environment.Finally, in the evoluIonary scenario of autonomous driving technologies, the transiIonal phase characterized by parIal vehicle autonomy presents criIcal challenges in human-machine interacIon [11].The emerging paradigm necessitates a sophisIcated, bidirecIonal monitoring system that transcends tradiIonal human-vehicle interfaces. In fact, the vehicle must develop robust capabiliIes for comprehensive driver state monitoring, including advanced sensor fusion techniques for enabling holisIc driver condiIon evaluaIon, to determine whether the driver is able to intervene and take control [12].Empirical research suggests that effecIve human-autonomous system collaboraIon requires not just technological sophisIcaIon, but a deep understanding of human cogniIve and physiological variability [13]. The ulImate goal is to develop a symbioIc human-machine interface where autonomous systems act as collaboraIve partners rather than mere technological subsItutes, enhancing overall transportaIon safety and efficiency. In this regard, in this arIcle collecIon, Giorgi and colleagues propose an integrated framework to assess the driver's mental faIgue in real Ime using electroencephalographic, electrooculographic, photoplethysmographic, and electrodermal acIvity. Their holisIc approach showed that the most sensiIve and Imely parameters are those related to brain acIvity. To a lesser extent, those related to ocular parameters are also sensiIve to the onset of mental faIgue but with a delayed effect.In conclusion, the confluence of neuromorphic compuIng, advanced sensing technologies, and sophisIcated human-machine interacIon represents a pivotal transformaIon in autonomous driving. This collecIon illuminates the mulIfaceted challenges and innovaIve soluIons emerging at the intersecIon of arIficial intelligence, neuroscience, and automoIve engineering. From energy-efficient neural controllers to adapIve percepIon systems and comprehensive driver state monitoring, the research demonstrates that autonomous driving is far more than a technological challenge-it is a complex socio-technical ecosystem.The future of autonomous vehicles might lie not in complete human replacement, but in creaIng intelligent, responsive systems that collaborate seamlessly with human operators. By integraIng biomimeIc computaIonal approaches, advanced sensor fusion, and a nuanced understanding of human cogniIve variability, we move closer to a transportaIon paradigm that prioriIzes safety, efficiency, and harmonious human-machine interacIon.