Electronic Architecture for Autonomous Vehicles

1. Introduction

The introduction of vehicles with a higher level of automation is proving to be a challenge, especially when dealing with complex scenarios in which sensory perception does not allow for complete and robust situational awareness, but in which the system must be able to make decisions. All of this is leading to a change in the electronic architecture, enabling the merging of information from different sensors and its processing in control units to offer more complete decision-making solutions. Therefore, it is crucial to study the design of this architecture to offer sufficient flexibility for the advances that are being introduced at the application level. This architecture includes the organization of sensors and control units. Aspects that guarantee the safety and the security of the system and its robustness must also be considered, as well as the relationship between the electronic architecture and embedded software, since the latter must be adapted to perform calculations in the most efficient way possible on large volumes of information that may be incomplete or not totally reliable. Furthermore, in addition to road vehicles, circulating on highways or urban environments, sectors such as the military, agriculture, construction, etc., may require specific architectures that optimize performance. Finally, the architecture must be open to different control strategies such as automated guidance, remote guidance, or remote monitoring, as well as to information sources that convey data from the vehicle’s own sensors to outside wireless communication systems (other vehicles or infrastructure).

With the aim of presenting key advancements in the field of electronic and logical architecture for highly automated vehicles, a Special Issue entitled “Electronic Architecture for Autonomous Vehicles” was proposed in Electronics. The call for papers invited contributions related to electronic and software architecture that must be implemented in highly automated vehicles so that they can deal with complex scenarios by processing the information obtained via on-board sensors and/or wireless communications. It also aims to present a variety of approaches that are currently being utilized, as well as the application of architectures oriented to certain sectors or employed in specific contexts. State-of-the-art studies were also welcome. This Special Issue ultimately brought together five accepted papers published in Electronics, which offer a broad perspective on the problem and some solutions that are currently being proposed.

2. Contributions

Within the architecture of automated vehicles, the perception and understanding of the environment—to form what is known as situational awareness—are essential, as they serve as the foundation for subsequent decision-making. In this context, the training of the systems used to interpret this environment is critical; however, these systems often struggle in safety-critical situations, precisely because such scenarios occur infrequently. Zhou et al. [Contribution 1] address the collection and annotation of video datasets featuring pedestrians in real-world environments—an approach that has proven insufficient for covering rare or safety-critical scenarios. Consequently, they explore “world foundation models,” which have demonstrated remarkable capabilities in generating realistic videos. Through quantitative evaluation and structured human assessment, the authors identify several challenges that must be resolved before videos generated by these models can be considered reliable for safety-critical autonomous driving applications.

Furthermore, Quezada-Godoy et al. [Contribution 2] propose a study on the behavior of automated vehicles in complex scenarios—such as intricate urban environments—where uncertainty in perception, dynamic interactions among multiple agents, and control instability under adverse conditions remain significant challenges. The study focuses on evaluating the robustness of a perception and decision-making architecture across five scenarios: heavy rain, dense fog, nighttime driving, heavy traffic, and a combination of extreme conditions.

Finally, at the actuation layer of automated vehicles, it is crucial to examine the integrity of the low-level electronic signals connecting controllers to actuators, to detect hardware degradation, timing issues, software faults, or unforeseen interactions, and to take appropriate corrective measures. Malayjerdi et al. [Contribution 3] propose a low-level anomaly detection layer, based on artificial intelligence, for autonomous vehicle architectures. The core concept involves integrating a lightweight observer into a virtualized master controller, capable of monitoring control signal flows in real time without interfering with the primary control logic.

In general, it should be noted that the electrical and electronic architecture of vehicles has evolved rapidly toward a centralized model. Currently, there is no unified industry consensus regarding the research strategy for these platforms. Zhang et al. [Contribution 4] employ a fuzzy-set qualitative comparative analysis method to identify viable combinations of R&D strategies that meet the requirements for architectural advancement. Furthermore, they develop a total-cost evaluation system, focusing on the software and hardware elements linked to the architecture. The study systematically analyzes combinations of R&D strategies that optimize costs for automotive companies. Similarly, Bousia [Contribution 5] conducts a systematic analysis of the state of the art regarding the use of business intelligence (BI), artificial intelligence (AI), data analytics, and machine learning in the transportation sector.

3. Conclusions

Despite significant advancements—such as the introduction of driver-assistance systems and the notable deployment of vehicles with intermediate levels of automation—highly automated vehicles still present major limitations and challenges that must be addressed. This Special Issue focuses on this unprecedented technological and social transformation, in which automated vehicles promise to evolve our concept of mobility, as well as reshape our cities, laws, and ways of life.

In this context, the issue includes contributions regarding solutions (and limitations) in areas such as environment perception and interpretation, decision-making in particularly complex scenarios—which currently constrain the operational scope of these vehicles—and the identification of faults for the implementation of countermeasures. Furthermore, it highlights the need to develop coherent and viable strategies to ensure the effective future deployment of these vehicles. Thus, this collection of papers provides an overview of the key pillars upon which research should be built, also illustrating in a specific manner the problems that must be tackled.