New Paradigms in Automotive Engineering

Abstract

Driven by global energy transformation and the progress of artificial intelligence technology, traditional automotive engineering is undergoing profound changes. Transportation is rapidly advancing toward electrification and intelligence. Against this background, this paper identifies three emerging paradigms for the development of electric vehicles: Heart Revolution, Brain Evolution, and Network Integration. This paper points out that automobiles are evolving from traditional one-way energy consumers to dynamic energy nodes in smart grids. With the support of artificial intelligence technology, the role of automobiles is also shifting from a simple means of transportation to an intelligent mobile terminal. At the same time, this paper focuses on analyzing the application of the integration theory of “Four Networks and Four Flows” in automobile upgrading. The theory does not focus on the optimization of a single node unit but emphasizes a systematic perspective to improve overall performance and support sustainable development. This paper suggests that the development of the automobile industry must be deeply integrated with the humanity world, information world and physical world. By building a five-in-one architecture of “Human–Vehicle–Road–Cloud–Satellite”, the automobile industry could follow a practical pathway toward coordinated development. At the same time, breakthroughs in core technologies such as solid-state batteries and wide-bandgap semiconductors are also imminent. This paper aims to provide a sustainable and high-performance automobile development path and integrate the concept of human-oriented design into it. Meanwhile, China’s new energy vehicle industry is used as a representative context to illustrate its engineering and industrial implementation.

1. Introduction

The global automobile industry is at a key turning point, evolving from traditional means of transportation to intelligent mobile terminals. This automobile revolution transcends the traditional technological upgrading and industrial iteration, representing a new change driven by value, as shown in Figure 1. With the electrification and intelligent development of transportation, the core power source of vehicles has shifted from traditional internal combustion engines plus gasoline or diesel to electric motors plus electric energy via batteries, which has significantly improved efficiency and power density. At the same time, the characteristics of the two-way transmission of electricity make electric vehicles (EVs) a dynamic mobile node in the smart grid [1]. Recent studies on vehicle-to-grid (V2G) integration further suggest that EVs can serve as controllable loads and distributed energy storage resources, supporting grid flexibility, renewable energy integration, and demand side coordination [2,3]. In addition, the rapid development of sensors, operating systems and artificial intelligence (AI) has given vehicles the ability to perceive, make decisions and reason [4]. The emergence of ChatGPT (GPT-3.5, OpenAI) in 2022 ignited the booming development of global artificial intelligence. So far, generative artificial intelligence has shown a diversified and innovative prospect. With the rapid iteration and development of artificial intelligence technology, cars will become more and more intelligent. Figure 2 shows the proportion of electric vehicle sales in countries worldwide in 2024 [5]. This shows that China and European countries are accelerating the development of the electric vehicle industry. This highlights that electric vehicles have entered the stage of scale and widespread popularization.

In order to cope with this far-reaching change, we must promote the deep integration of the humanity world, the information world and the physical world. The progress of the automobile industry and AI technology depends on the practical application scenarios in the field of physics. The development of artificial intelligence engines is based on the philosophical concept of deep integration of the humanity world, the physical world, and the information world, pursuing the value orientation of human–machine collaboration and virtual–real symbiosis. In terms of scientific thought, it focuses on revealing and understanding the three core paradigms of large language models, digital twins, and big data, and grasping the internal operating mechanism of intelligent systems; in terms of engineering thought, it integrates and coordinates the three paradigms through technological implementation and system optimization, transforming scientific cognition into stable, efficient, and implementable capabilities, ultimately generating practical applications and social value. In addition, the integration of AI experts and scenario experts is an important dual engine for AI to move from technological hype to the deep waters of industry. AI experts understand technology, models, and patterns, while scenario experts understand business, pain points, and value. Deep collaboration between the two is essential for technology to truly be implemented, solve real-world problems, and create industrial value. The future’s initiative lies not in the hands of a single technology enabler but in those who proactively integrate AI with specific scenarios. To be specific, AI technology has accelerated the integration of automotive systems, and the attributes of automotive systems have changed, as shown in Figure 3.

At the same time, their development must be constrained by policies, standards and user needs in the field of humanities, which are rooted in the concept of human-oriented design. The “Four Networks and Four Flows (4N4F)” integration theory (four networks include energy network, information network, transportation network and humanities network; four flows include energy flow, information flow, material flow and value flow), proposed by C. C. Chan [6,7], provides a specific operating framework for the development of the automobile industry, as shown in Figure 4. The theory unifies the previously independent social units into a coherent theoretical system, thus optimizing the performance of the whole system.

In this study, the 4N4F framework is a way of thinking to understand the problem: it helps us see how the car has gradually transformed from a simple means of transportation into an interconnected part of energy, information, transportation, and even social systems. This framework has its scope and limitations. First, 4N4F is not a rigorously data-validated engineering model, nor is it a universally applicable predictive theory. It simply helps us clarify and explain the interactions between multiple systems during the automotive transformation process. Second, the situations vary greatly from place to place: energy and digital infrastructure, transportation systems, policies and regulations, market maturity, and user habits all differ. Therefore, the applicability of this framework will vary from place to place and cannot be applied indiscriminately.

This article will take the 4N4F integration theory as the perspective and deeply explore the opportunities, challenges and actionable guidance brought about by the transformation of the automobile industry. The research framework and roadmap of this article are shown in Figure 5. The three paradigms, the 4N4F framework, and the Human–Vehicle–Road–Cloud–Satellite (HVRCS) architecture are not independent concepts but rather three representations of the same transformation at different levels. The three paradigms describe the fundamental principles of automotive evolution: the First Paradigm mainly corresponds to the energy network and energy flow, the Second Paradigm mainly corresponds to the information network and information flow, and the Third Paradigm describes the coordinated operation of all four networks and four flows. The 4N4F framework thus provides the system-level analytical model, while HVRCS provides the engineering implementation architecture. Within this mapping, the vehicle is the central mobile node; the road belongs primarily to the transportation network but also supports information and energy flows; the cloud and satellite belong primarily to the information network; and human corresponds to the humanistic network. Accordingly, AI computing is positioned mainly in the information network, while AI-generated data are treated primarily as information flow and secondarily as value flow when converted into system optimization and social or economic benefits. Based on this logic, this paper is organized as follows. First, it summarizes the new paradigms of automotive engineering under the current industrial transformation. It then examines the evolution of AI-enabled in-vehicle computing, followed by a discussion of the HVRCS system as the implementation architecture for cross-domain integration. China’s new energy vehicle industry provides a representative context for discussing how intelligent electric vehicles may be further developed through the coordinated advancement of AI models, in-vehicle computing architectures, and automotive chips. Finally, it considers the future energy ecosystem within the 4N4F framework and highlights key technologies such as solid-state batteries, high-efficiency electric drive systems, and wide-bandgap semiconductors. This article aims to provide technical guidance for building a sustainable, high-performance, and human-centered future automobile industry.

2. New Paradigms of Automotive Engineering

At the intersection of the fourth and fifth industrial revolutions, automobiles, as the core products of industrial civilization, are undergoing a fundamental change. They are evolving from single-function vehicles to mobile intelligent terminals that integrate energy, information, computing and human-oriented design. This change is not just an improvement in performance or an iteration of technology but represents a new engineering paradigm that reshapes the “heart” and “brain” of the car, as shown in Figure 6. The integration theory of 4N4F shows that automobiles are no longer an isolated mechanical system but an intelligent terminal with perception, decision-making and collaboration capabilities, and are deeply integrated with energy, information and social systems.

2.1. The First Paradigm: The “Heart” Revolution

For a long time, the internal combustion engine, which is the “heart” of traditional vehicles, has not been able to shed the thermodynamic limitations of the Carnot cycle [8]. Its theoretical upper limit of thermal efficiency is defined by the Carnot cycle efficiency formula:

$$W=Q_1-T_2{Q}_1{\displaystyle \frac{1}{T_1}}\Rightarrow \eta ={\displaystyle \frac{W}{Q_1}}=1-{\displaystyle \frac{T_L}{T_H}}$$

(1)

where T_H is the high-temperature heat source (the highest temperature of the combustion gas inside the cylinder), and T_L is the low-temperature heat source (usually the ambient air temperature). Due to the limitations of material heat resistance limits and ambient exhaust temperature, the effective thermal efficiency of traditional internal combustion engines has long hovered between 30% and 45% [9]. In the process of converting chemical energy into kinetic energy, a large amount of energy is wasted in the form of heat loss and carbon emissions. In contrast, electric vehicles use electric motors to convert power, achieving high efficiency (more than 90%) [10]. In addition, electric motors are also superior to internal combustion engines in terms of response speed and power density. The transformation from internal combustion engine to electric motor represents a huge change in the way primary energy is used. Although the energy density of power batteries currently lags behind that of liquid fuels, the electric drive system has broken through the limitations of traditional power systems in terms of system efficiency, control accuracy and environmental friendliness. At the same time, with the continuous improvement and breakthrough of the performance of the power battery, its energy density and safety will also be rapidly improved.

The most basic feature of electric vehicles is their energy carrier-electricity. It realizes two-way energy flow. In contrast, the fuel system of traditional cars involves liquid fuel passing through the internal combustion engine, resulting in one-way energy flow. However, in the vehicle-to-grid (V2G) framework [11], electric vehicles are defined as dynamic nodes in the energy network. As shown in Figure 7, they use two-way wired or wireless charging devices to promote two-way power flow according to the power grid and user needs. Real-world pilots have validated this architecture. The Parker Project in Denmark demonstrated that a fleet of EVs could deliver measurable frequency regulation services to the national grid [12], while the Utrecht vehicle-to-grid deployment in the Netherlands showed that bidirectional charging across hundreds of vehicles could offset peak grid demand without degrading battery performance [13].

2.2. The Second Paradigm: The “Brain” Evolution

If the power system is the “blood circulation system” of electric vehicles, then the computing system with high-performance chips as the core, built by the operating system and driven by AI algorithms, constitutes the “brain” of the vehicle. This intelligent center gives cars the ability to perceive and reason. With this “brain”, vehicles no longer rely only on the driver’s instructions. On the contrary, it can perceive the surrounding environment, analyze data, and assist drivers in making decisions through sensors, operating systems and artificial intelligence algorithms. This “brain” can also learn the driver’s driving habits and dynamically optimize the power response, energy consumption relationship and other factors according to different usage scenarios. The accumulation of this experience gives electric vehicles the characteristics of personalization.

The next generation of intelligent vehicles is gradually establishing a highly coupled operating system of perception, decision-making, and execution. Multimodal sensors and vehicle-to-infrastructure information form the foundation for real-time perception of the physical world, high-performance computing platforms and AI models support rapid judgment in complex scenarios, and chassis-by-wire technology accurately translates decision results into vehicle behavior. In practice, Honda’s SENSING Elite system, commercially deployed in Japan in 2021, demonstrated L3 autonomous operation in congested highway conditions [14], providing early evidence that the perception–decision–execution loop described here is achievable at scale.

In addition, the digital twin models consist of semi-physical simulation and optimization [15]. They map system data and provide input for optimization and enable continuous iteration of algorithms and models based on data feedback, driving the continuous evolution of vehicle performance through interaction between the virtual and real worlds.

2.3. The Third Paradigm: The “Network” Integration

The automotive revolution cannot happen in isolation. Figure 8 shows the schematic diagram of the integration of the new energy vehicle industry. The development of new energy vehicles depends on the coordinated integration of upstream components, energy infrastructure, digital infrastructure, smart traffic, and smart cities. Future automobiles should be considered as key nodes within a broader industrial and urban ecosystem. Only when automotive engineering is systematically coupled with the energy revolution, the information revolution, and social development goals can its engineering potential and social value be truly unleashed.

With the rapid development of technologies such as artificial intelligence and the Internet of Things, automobiles are no longer just a simple collection of mechanical, electrical or control engineering but are embedded in a complex system covering energy, transportation, information and human behavior, as shown in Figure 9.

According to the 4N4F integration theory, the energy network, information network, transportation network and humanistic network together form the basic framework for the operation of human society. The energy network is the physical foundation of social operation. At present, the energy network is undergoing significant changes, from a centralized system to a distributed system, and from a single fossil fuel system to a multi-source new energy system. The information network accelerates the dissemination, integration and computing speed of information by deploying multiple sensors and supercomputing capabilities. At the same time, the transportation network has continuously improved its flexibility through electrification and intelligent upgrading and has become an important channel for multi-network coupling. Guided by the principle of human-oriented development, the humanistic network defines operational boundaries and development goals through institutional frameworks, codes of conduct and value constraints.

Corresponding to the structural coupling of the “four networks,” the “four flows” reveal the internal mechanisms of system operation. Energy flow, material flow, and information flow interact within the system. Through the scheduling and modeling of information flow, energy and materials are no longer merely passively consumed objects but become system resources that can be optimized and scheduled. On this basis, value flow gradually emerges as the result of the collaborative operation of the system, demonstrating a unity of economic, social, and ecological benefits.

From the perspective of the integration of 4N4F, the engineering positioning of automobiles has undergone a huge change. Automobiles are no longer just independent terminals in the transportation network but have gradually evolved into mobile energy storage and regulation nodes in the energy network, edge sensing and computing nodes in the information network, and important execution units in the value flow. Therefore, the new rules of automotive engineering are no longer defined by a single performance indicator but depend on their collaborative capabilities and overall value contribution at the system level.

2.4. The Future Form of the Automobile: A Super-Intelligent Lifeform

Within the framework of the 4N4F integration, the engineering connotation of future automobiles is no longer limited to the continuous optimization of transportation functions, but points to a completely new system form—a mobile intelligent space integrating the human world, the information world, and the physical world. This system form can be understood as the future outcome of the three emerging paradigms discussed above: the Heart Revolution provides the vehicle with a new energy circulation capability, the Brain Evolution enables perception, computation, learning, and decision-making, and Network Integration embeds the vehicle into broader energy, information, transportation, and humanity networks. Therefore, automobiles are evolving from traditional means of transportation into super-intelligent life forms, which serve as a conceptual summary of next-generation automobiles under the combined support of the three paradigms and the 4N4F framework. These future automobiles will be capable of sensing the environment, understanding users, and participating in the collaborative operation of the system, as shown in

Table 1. Strategic framework: the future automotive super-intelligent lifeform.

Core Characteristic	Technical Definition & Morphology	Key Enabling Elements	Strategic Objective	Representative Use Cases
Deep Integration	A mobile smart space merging the human, information, and physical worlds.	4N4F: Interconnection of Energy, Information, Transportation, and Human networks.	Link the automotive revolution with energy and information revolutions.	Smart charging infrastructure
Autonomous Evolution	A complete intelligent closed loop of perception, decision-making, and execution.	AI Deep Learning: Continuous growth and self-improvement through data iteration.	Create a vehicle that grows and learns from its environment.	Autonomous driving function updates
Ecological Symbiosis	Functions as a critical node in the smart energy system.	V2G: Bi-directional interaction to participate in peak shaving and energy storage.	Achieve harmony between the vehicle and the broader energy ecosystem.	Peak shaving and valley filling through V2G
Human-Centric Interaction	Focused on user experience to build emotional bonds.	Emotional & Personalized Interaction: Tailored services that understand user intent and feelings.	Transform the vehicle from a tool into an empathetic smart partner.	Privacy-aware user services.

.

Firstly, the most essential characteristic of a super-intelligent life form lies in its deep integration. Future automobiles will no longer operate within a single transportation system but will be simultaneously embedded in a composite system consisting of energy networks, information networks, transportation networks, and human networks. Within this system, energy flow, information flow, material flow, and value flow will no longer be isolated but will achieve collaborative optimization at the system level. The automobile is both an important node in the energy flow and an important carrier of the information flow, simultaneously undertaking the dual roles of material movement and value creation.

Secondly, self-star transformation is the key ability to distinguish ultra-intelligent life forms from traditional vehicles. Based on the deep integration of high-performance computing platforms, operating systems and artificial intelligence algorithms, future vehicles will gradually build a highly coupled intelligent closed-loop control system, covering three aspects of perception, decision-making and execution. Vehicles use multi-sensor systems to perceive the external environment and optimize decision-making through artificial intelligence systems. The whole process is autonomously carried out by on-board intelligent systems, which continuously improve algorithms by learning the driver’s driving habits so as to continuously improve vehicle performance and user experience.

Thirdly, ecological symbiosis is the key embodiment of the integration of vehicles into the social system in the future. Under the background of traffic electrification and intelligence, electric vehicles can use V2G technology to realize two-way power flow between the power grid and vehicles. This not only meets the travel needs of human beings but also contributes to the stable operation of the power grid. By balancing the development of vehicles, energy systems and urban infrastructure, the sustainable development of cities can be promoted.

Ultimately, putting humanity first is the core value of this super-intelligent life form. Future automotive optimization will be driven by human needs, significantly enhancing the user experience. Thanks to intelligent cockpits, personalized services, and emotional interaction mechanisms, automobiles will gradually move from the status of impersonal technical products to that of mobile spaces with attributes enabling a more natural, intuitive, and personalized interaction between humans and vehicles.

3. Impacts and Applications

3.1. Development Path of Automotive AI Models, In-Vehicle Computing Power Evolution and Domestic Chip Development Opportunities

The in-depth integration of automotive technology and AI is the core driving force for promoting the large-scale implementation of autonomous driving and reconstructing the smart mobility ecosystem. Different from technological exploration in general AI scenarios, the development of AI models in automotive scenarios must be based on the goal of engineering implementation, with solving actual driving needs, improving user experience, and achieving commercialization and industrialization as the core orientation so as to avoid disconnection between technology and the market. Combining industry development trends and technological iteration trends, the development of AI models in automotive scenarios should follow the three-step path of perceptual intelligence, cognitive intelligence, general intelligence, synchronously support the upgrading of in-vehicle computing power and the evolution of architecture. On this basis, this study further focuses on China’s industrial strategic layout and explores the development opportunities of domestic core automotive chips so as to provide targeted industrial development references [17]. The following will elaborate on the development path of AI models, the current situation, and future trends of in-vehicle computing power, and put forward targeted suggestions based on actual needs.

(1)

Development Path of AI Models in Automotive Scenarios

The iteration of AI models in automotive scenarios is essentially a process of gradually realizing that machines are replacing humans in completing driving tasks. The development rhythm of AI is deeply bound to the evolution of autonomous driving levels (L2–L5). Each stage is guided by clear engineering implementation goals, focusing on core technological breakthroughs while considering user needs and market acceptance, and it is promoted step by step in three specific stages. The classification and core features of vehicle driving automation levels are shown in

Table 2. Classification and core features of vehicle driving automation level.

Level	Official Name	Core Definition	Key Features
0	Emergency Assistance	No sustained lateral/longitudinal vehicle control; only partial OEDR capability	Driver holds full responsibility; only provides warning & momentary emergency intervention
1	Partial Driver Assistance	Sustained lateral OR longitudinal motion control, with matched partial OEDR within ODC	Driver holds full responsibility; system controls single motion axis; driver monitors full-time
2	Combined Driver Assistance	Sustained lateral and longitudinal motion control, with matched partial OEDR within ODC	Driver holds full responsibility; system controls both motion axes; driver must monitor & take over anytime
3	Conditionally Automated Driving	Sustained full DDT execution within defined ODC	System is responsible within ODD; user only needs to respond to takeover request in time
4	Highly Automated Driving	Sustained full DDT execution + automatic MRM within defined ODC	System holds full responsibility within ODD; no user takeover obligation; executes MRM automatically
5	Fully Automated Driving	Sustained full DDT execution + automatic MRM under all drivable conditions	No ODD limit; system holds full responsibility; no takeover requirement

Note: OEDR: Object and Event Detection and Response; ODC: Operational Design Condition; ODD: Operational Design Domain; DDT: Dynamic Driving Task; MRM: Minimal Risk Maneuver.

[18].

The first phase (from now to 2027) focuses on perceptual intelligence to achieve large-scale implementation of L3-level autonomous driving. This stage is the foundation-laying period for AI models in automotive scenarios. The core goal is to break through the bottlenecks of perception accuracy and real-time performance, complete the technical verification and commercialization of L3-level autonomous driving, and, at the same time, empower intelligent cockpits relying on large AI models to improve user interaction experience. Currently, the application of automotive AI models is mainly concentrated in the perception layer and the basic interaction layer. The core task is to achieve accurate identification and real-time perception of driving environment elements through multi-sensor fusion, while supporting the stable operation of L2+ level assisted driving functions. In terms of technical implementation, this stage needs to focus on solving the robustness problem of perception models and improving the accuracy of environmental recognition; meanwhile, with the help of the scenario understanding ability of large AI models, empower functions such as voice interaction and cockpit scenario-based services, and break the interaction barriers of traditional cockpits.

The second phase (2027–2030) moves towards cognitive intelligence, achieving a breakthrough in the commercialization of L4-level autonomous driving. This phase is an upgrade period for AI models in automotive scenarios, with the core goal of transitioning from perception to cognition. Relying on end-to-end large models, it will achieve a breakthrough in the commercialization of L4-level autonomous driving, enabling vehicles to have reasoning and decision-making capabilities in complex scenarios and truly understand the driving environment and user intentions. Compared with the perceptual intelligence of the first phase, cognitive intelligence requires AI models not only to see the driving environment but also to analyze scene logic, judge potential risks, and decide the optimal driving plan. The core technological breakthrough in this phase is the end-to-end autonomous driving large model, which abandons the traditional modular design approach. It directly outputs vehicle control commands through a single neural network, realizing the integration from perception to decision-making and control, and greatly improving the adaptability and decision-making efficiency in complex scenarios. At the same time, AI models will achieve improved generalization capabilities across scenarios, gradually covering full-scenario driving tasks; the collaboration between intelligent cockpits and autonomous driving will also be further deepened. Relevant studies have shown that the integration of a Large Language Model (LLM) can significantly improve the reasoning and decision-making capabilities of the model [19].

The third phase (after 2030) realizes general intelligence and creates an intelligent travel partner. This phase is a mature stage for AI models in automotive scenarios, with the core goal of achieving general intelligence, making vehicles intelligent partners with autonomous learning, emotional interaction, and multi-scenario adaptation capabilities, completely liberating human drivers, and realizing the popularization of full-scenario unmanned driving. At this time, AI models will have autonomous learning capabilities, which can continuously optimize model performance through daily driving data; at the same time, emotional interaction capabilities will become a core competitive advantage, and the model can provide personalized emotional care and services by analyzing user status. In addition, AI models will break through the limitations of a single vehicle, realize the global collaboration of vehicles, roads, people, and clouds, and become a core node in the intelligent transportation ecosystem. The technical focus of this phase is the in-depth integration of Artificial General Intelligence (AGI) with automotive scenarios, with core breakthroughs in autonomous learning algorithms, emotional computing models, and global collaborative intelligent models, ultimately realizing the autonomous collaboration of human-vehicle-environment [20].

Figure 10 clearly presents the development path of AI models, and combines the core goals, key technologies, and computing power requirements.

(2)

Current Situation and Development Trends of Vehicle-Side Computing Power

The implementation and iteration of AI models in automotive scenarios are inseparable from the support of vehicle-side computing power. As AI models evolve from perceptual intelligence to cognitive intelligence and general intelligence, the demand for vehicle-side computing power shows exponential growth. At the same time, the computing power architecture and collaboration model are also undergoing changes, gradually evolving from distributed to centralized, and from a single vehicle-side to end-cloud collaboration, providing a guarantee for the engineering implementation of AI models.

Currently, the development of in-vehicle computing power is in a critical stage of transitioning from meeting basic perception needs to supporting high-level cognitive needs. Its core feature is that computing power requirements are deeply tied to autonomous driving levels, and the computing power architecture is dominated by domain controllers.

In terms of computing power requirements, each upgrade in the autonomous driving level leads to an approximate order-of-magnitude increase in computing power demand: L2-level assisted driving requires about 10 Tera Operations Per Second (TOPS); L5 level and future super intelligent entities will have a computing power demand exceeding 1000 TOPS, reaching the level of thousands of TOPS. Among them, TOPS is the core indicator for measuring the performance of in-vehicle computing power [21].

In terms of computing power architecture, the current mainstream is the distributed domain controller architecture, which divides the entire vehicle into multiple independent domains, such as power, cockpit, and autonomous driving. This architecture has low development difficulty and controllable costs, but it has drawbacks such as an inability to share computing power resources, low coordination efficiency among domains, and high hardware–software coupling, making it difficult to meet the needs of high-level autonomous driving and multi-model collaboration.

Combined with the development path of AI models and the needs of engineering implementation, the future in-vehicle computing power will present four core trends: exponential growth of computing power, centralized evolution of architecture, an end-cloud collaborative model, and breakthroughs in chip localization. The trend is demonstrated in Figure 11.

Trend 1: The demand for computing power continues to grow exponentially, and the energy efficiency ratio becomes the core competitiveness. With the improvement of model complexity and the surge in sensor data volume, the demand for in-vehicle computing power will continue to break through. L5 level requires 1000+ TOPS, and future super-intelligent organisms will require thousands of TOPS. At the same time, the vehicle environment has strict restrictions on power consumption, so the energy efficiency ratio will become the core competitiveness. To quantify the operational efficiency of an on-board computing platform, the AI energy efficiency ratio η_AI is defined as follows:

$${\eta }_{AI}={\varphi }_{comp}/P_{total}$$

(2)

where ϕ_comp represents the nominal computing power of the system (in TOPS), and P_total denotes the total power consumption of the processor under full-load conditions (in watts). Under limited power budgets, heterogeneous computing architectures (integrating CPUs, GPUs, and NPUs) can significantly enhance η_AI. This optimization is essential to provide stable and sustainable support for high-level autonomous driving functions while maintaining the vehicle’s overall energy economy.

Trend 2: The computing power architecture evolves from distributed to centralized [22]. In the future, it will evolve towards a central computing platform, which will uniformly handle all tasks, such as autonomous driving, intelligent cockpit, and body control, through one platform, realizing the sharing and efficient scheduling of computing power resources. The central computing platform will promote the decoupling of software and hardware, adopt a layered architecture, realize hardware replaceability and software upgradability, and support the rapid OTA iteration of AI models, which is the core direction of the era of software-defined automobiles.

Trend 3: Edge computing and cloud collaboration deepen, building a cloud-edge-end integrated computing power system. In the future, in-vehicle computing power will be deeply coordinated with cloud and edge computing power: the vehicle end (edge end) is responsible for processing perception, decision-making and control tasks with high real-time requirements; the cloud is responsible for processing high-complexity non-real-time tasks such as model training and big data analysis; edge nodes (such as roadside equipment) supplement the vehicle-end computing power to improve safety and generalization ability. This collaborative model can balance computing power costs and real-time requirements and support the continuous optimization of models, which is an important support for the evolution of AI towards general intelligence.

(3)

Development Opportunities for Domestic Chips

Considering the development path of AI models in automotive scenarios and the evolution trend of on-vehicle computing power, focusing on the goals of engineering implementation and industrialization development, we should seize the historic opportunity of domestic chips, promote the collaborative breakthrough of technology and industry, and realize the closed-loop development of AI models, computing power architecture, and chips.

China’s new energy vehicles have achieved overtaking on a curve. As the core support for the implementation of AI models, the independent and controllable nature of on-vehicle computing power chips has become the key to the high-quality development of the automotive industry [23]. If chips cannot achieve synchronous breakthroughs, this will restrict the implementation of AI models and autonomous driving technologies and affect the industrial competitiveness of China’s new energy vehicles. Therefore, domestic chips must seize this historic opportunity, develop synchronously with the new energy vehicle industry, and achieve independent control of core technologies. Combined with the development trend of on-vehicle computing power, it is recommended to focus on three development directions.

Firstly, focus on automotive-grade AI chips, benchmark against international advanced levels, and break through core technical bottlenecks. Automotive-grade AI chips are the core of on-vehicle computing power, directly determining the operational performance and real-time nature of AI models. Currently, domestic automotive-grade AI chips have been mass-produced and implemented in advanced manufacturing processes. However, the high-end market still relies on international manufacturers, and there are still shortcomings in key links such as EDA tools, IP cores, and functional safety certification. In the future, it is necessary to increase R&D investment, break through core technologies such as heterogeneous computing architecture, low-power design, and functional safety, benchmark against international advanced levels, and achieve independent control of high-end chips at L4–L5 levels. At the same time, it is necessary to combine the needs of domestic automakers to develop customized chips, improve adaptability with vehicle architectures, and reduce application costs.

Secondly, make efforts in power semiconductors, focusing on breaking through third-generation wide-bandgap semiconductor power devices. Power semiconductors are the core components of on-vehicle computing power chips, power batteries, and motor control, directly affecting vehicle energy consumption, range, and computing power stability. Third-generation wide-bandgap semiconductors (such as SiC, GaN) have the advantages of high temperature resistance, high voltage resistance, low power consumption, and high efficiency, which can significantly improve the energy efficiency ratio of on-vehicle computing power chips and are the core development direction in the future. Currently, this technology is still in its developing stage, and domestic enterprises still have gaps in materials, manufacturing, and other links. It is necessary to increase R&D investment, break through core technologies, achieve large-scale mass production, and support industrial upgrading and development [24].

Thirdly, lay out intelligent sensor chips to improve the perception end and guarantee computing power support. Intelligent sensors are the eyes of AI models, and their data accuracy directly determines the perception and decision-making accuracy of the models. With the popularization of multi-sensor fusion technology, higher requirements have been put forward for the accuracy, power consumption, and integration of intelligent sensor chips. In the future, it will be necessary to focus on the research and development of intelligent sensor chips and breakthrough technologies, such as high-precision perception, low-power integration, and multi-modal fusion while at the same time promoting the collaborative design of sensor chips and automotive-grade AI chips to achieve integration of perception-computing power-decision-making.

3.2. A Nationally Aligned Technical Pathway for the Coordinated Development of Human–Vehicle–Road–Cloud–Satellite Systems

The automobile as a super-intelligent lifeform represents the future form of intelligent electric vehicles under the 4N4F framework. In this section, the Human–Vehicle–Road–Cloud–Satellite system is further discussed as a practical engineering pathway for realizing this vision, with the vehicle serving as the core mobile carrier. The transition of automotive engineering into an intelligent, connected, and system-oriented discipline requires a coherent technical pathway that aligns engineering design with industrial development objectives. In this context, the coordinated development of HVRCS systems provides a scalable engineering route for next-generation mobility systems. Rather than treating humans, vehicles, infrastructure, digital platforms, and spatial communication systems as isolated subsystems, this pathway emphasizes their deep coupling and co-evolution under unified architectural principles and governance frameworks. An illustration of the technical pathway is shown in Figure 12. In fact, the integration is mainly reflected in two aspects: first, the deep integration of automobiles, energy, and information, making intelligent electric vehicles the core carrier; second, the full interconnection of human, vehicle, road, cloud, and satellite, promoting the industry to gradually evolve from intelligent electric vehicles to intelligent transportation and smart cities, and ultimately building a smart society, as shown in Figure 13.

The core of this technical path is the human-centered engineering concept. The industrial strategy pays more and more attention to the safety, user experience and social value of intelligent transportation systems, thus reshaping the design goals of automotive engineering [25]. Human beings are no longer external users who interact with vehicles but are an inherent component of the travel system. Their behavior, preferences and cognitive limitations must be systematically modeled and incorporated into engineering design [26]. This requires the transformation of traditional human–computer interaction into human–computer system integration. The process of perception, decision-making and control needs to be jointly optimized between human and machine subjects. By embedding human intention identification, behavior prediction and trust perception control mechanisms into vehicles and platforms, the HVRCS system ensures that technological intelligence is consistent with travel needs and social expectations in the real world.

In this mode, vehicles operate not only as transportation tools but also as intelligent network entities that connect the human world with the physical and information domains.

Table 3. Comparative analysis of advanced driver assistance systems hardware architectures across global automakers.

Model	Assisted Driving Level	No. of Cameras	No. of Radars	No. of LiDAR	Computing Power
Tesla (Model S/3/Y/CT)	L2 (Supervised FSD v12)	8	0	0	Est. 300–500 TOPS
Mercedes-Benz (S-Class/EQS)	L3 (Drive Pilot)	7	5	1	Varies by platform
BMW (7 Series/i7)	L2+/L3 (Personal Pilot)	6 to 7	5	1	Varies (Qualcomm Ride)
NIO (ET7/NT2.0)	L2+/L3-ready	11	5	1	1016 TOPS
XPeng (G9/X9/P7+)	L2+ (XNGP)	11	3 to 5	0 to 2	254 to 508 TOPS
Li Auto (L9/MEGA)	L2+ (End-to-End VLM)	11	1	0 or 1	128 to 508 TOPS
Huawei (AITO M9/Luxeed)	L2+ (Qiankun ADS 3.0)	11	3	1	~200–400+ TOPS

summarizes representative electric vehicle brands and models, together with their assisted-driving level, sensor configuration, and onboard computing power. As the table shows, mainstream electric vehicles are increasingly equipped with multiple cameras, radars, LiDAR sensors, and high-performance computing platforms. This reflects the rapid convergence of electrification, intelligence, and connectivity in modern vehicle design.

With this integration, vehicles are no longer independent mechanical products; instead, they are becoming mobile nodes in a distributed cyber–physical system. From a technical perspective, a vehicle simultaneously serves as an edge computing unit, a sensing platform, and an energy node. Its onboard perception and decision-making capabilities enable real-time responsiveness to dynamic driving environments, while standardized communication interfaces allow the vehicle to exchange data, status, and intentions with roads, cloud platforms, and other vehicles. This architectural transformation shifts intelligence away from being confined to a single vehicle and distributes it across the broader system. As a result, the overall transportation network gains improved scalability, redundancy, and cost-effectiveness. Therefore, vehicle performance and safety increasingly depend not only on isolated onboard functions but also on the shared intelligence enabled by interconnected sensing, computation, and communication infrastructures.

Based on the structural expansion and direction of the project, road infrastructure is the key to the gradual evolution from passive traffic carriers to intelligent collaborative system components. Investment in intelligent transportation infrastructure is a strategic solution-oriented approach, which recognizes that vehicle intelligence is not enough to ensure safety and efficiency in complex traffic environments [27,28]. By applying sensing, communication and dynamic wireless charging technology in road infrastructure, the HVRCS system provides collaborative perception and traffic-based coordinated energy control at the traffic system level [29]. Intelligent roads provide real-time information on traffic conditions, environmental factors and infrastructure status, enhance vehicle perception, and reduce uncertainty in extreme situations such as traffic jams, bad weather and mixed traffic flow. Road intelligence is a distributed security layer, which enhances the reliability of the whole system and promotes the gradual transition of the system to automation.

These theoretical benefits are increasingly validated in practice. Baidu’s Apollo Go service, operating fully driverless in Wuhan since 2022 across a V2X-equipped road network, provides one of the most extensive real-world datasets for HVRCS co-evolution [30]. In parallel, China’s Xiong’an intelligent road pilot has deployed roadside sensing units on over 100 km of infrastructure, enabling cooperative perception between vehicles and roads and reporting measurable reductions in collision risk at monitored intersections [31].

As the central analysis and coordination hub in the HVRCS technology path, the cloud platform emphasizes digital infrastructure and data-driven governance. In addition, the cloud platform can collect massive data from vehicles, roads and users. This will achieve continuous learning, algorithm iteration and system-level optimization, which cannot be achieved at the level of a single vehicle or infrastructure. Cloud-based insights can realize fleet coordination, traffic flow optimization, predictive maintenance, etc., so that the travel system can change from passive response to active attack. To achieve this proactive management mathematically, the cloud platform acts as a centralized solver that continuously minimizes a global system cost function, J. For a collaborative network of N vehicles, this objective function dynamically balances individual travel time, T_i, and energy consumption, E_i.

$$\min J={\displaystyle \sum _{i=1}^N\left({\alpha }_i{T}_i+{\beta }_i{E}_i\right)}+\gamma C_{infra}$$

(3)

where α_i and β_i are adaptive weighting factors for time and energy efficiency, and C_infra represents the operational state and cost of the infrastructure. Moreover, because these variables are now mathematically coupled, the cloud platform can integrate the transportation system with energy management, urban planning and public services, thus building an integrated Mobility-as-a-Service ecosystem. Thanks to cloud-edge collaboration, real-time decision-making can be localized, and long-term intelligence can be improved through centralized analysis and global optimization.

The satellite system can expand the coverage and time span of the HVRCS framework, providing key capabilities for national and cross-regional applications. High-precision positioning, time synchronization and satellite communication are important to ensure continuous service in areas where the ground network is difficult or cannot cover an entire area. At the industrial level, satellite integration is conducive to regional balanced development, infrastructure resilience and strategic security. The light satellite system helps to maintain redundancy and reliability technically so as to achieve consistent positioning and communication across multiple environments [32]. Through satellite data that works seamlessly with ground vehicles, roads and cloud systems, the HVRCS architecture provides the coverage and coordination capabilities necessary for large-scale intelligent traffic deployment.

The collaborative development of HVRCS systems is not purely based on the consideration of technological progress but is oriented by industrial applications. As an experimental environment, these intelligent interconnected vehicle demonstration areas can verify and improve the technical structure, standards and governance mechanism. These areas help to gradually integrate new technologies into practical application scenarios while controlling safety, supervision and social risks. In this regard, standardization has become one of the key driving factors, which can achieve interoperability between heterogeneous systems and stakeholders, and promote industrial cooperation and ecosystem development. The combination of open standards and modular architecture enables automobile manufacturers, information and communication technology providers, transportation operators and energy companies to carry out cooperative innovation within a common framework.

From the perspective of system engineering, the HVRCS technology path realizes the integration of the four networks of transportation network, information network, energy network and human network, as well as the corresponding material, information, energy and value flow. The information flow generated by vehicles and infrastructure provides information for cloud-based intelligent systems, thus optimizing the traffic and energy flow of the whole system. Energy flows support vehicle electrification and infrastructure operation, while value flows give rise to new business models and services. These flows are tightly coupled and dynamically regulated, resulting in emerging system behaviors, which cannot be fully understood or controlled by traditional vehicle-centered engineering methods. Therefore, in the case of strong cross-domain coupling and nonlinear interaction, new engineering paradigms are needed to manage the stability, security and performance of the management system.

In the end, the unified design of the HVRCS system is the result of the integration of the human world, the information world and the physical world. Therefore, automobile engineering has become the intersection of these fields, coordinating technological innovation with social values and national development goals. By matching the engineering design with the national industrial direction, the development path of HVRCS technology constitutes the cornerstone of automotive engineering a system that emphasizes collaboration rather than isolation and systems rather than components and is human-oriented rather than being based purely technical performance. This paradigm not only releases the transformation potential of intelligent travel but also sets a new set of rules for the conception, evaluation and management of future automobile systems.

3.3. Toward the Energy Revolution: Key Technologies for Mobility Electrification and Future Energy Ecosystem Frameworks

Under the “4N4F” framework, the energy network is a key component connecting transportation, information and industrial systems. From this perspective, the progress of the automobile industry does not occur in isolation but is synchronized with the broader energy transformation, which is redefining the way energy is produced, converted and used. The future energy system is developing towards decarbonization and electrification, and electric vehicles must evolve accordingly.

The energy transition also imposes new requirements on mobility electrification, including low-carbon operation, high energy conversion efficiency, flexibility in energy interaction, and compatibility with renewable energy sources. To meet these requirements, vehicle level energy storage and energy conversion technologies must be developed in a coordinated manner. Therefore, this section reviews four key technology directions: high-energy-density batteries, high-efficiency electric drive systems, wide-bandgap power electronics, and intelligent thermal management. Recent studies on V2G integration and automotive battery systems further indicate that large-scale mobility electrification requires coordinated development of vehicle-side technologies, charging infrastructure, grid flexibility, and lifecycle battery management.

At the same time, the electrification of transportation is also constantly reshaping the energy system. Renewable energy-based photovoltaic generation, stationary energy storage, charging facilities, and power-exchange infrastructure should be designed together with vehicle side charging and energy interfaces to improve system efficiency and scalability. In emerging power systems, V2G technology enables electric vehicles to act as distributed energy storage resources and supports the flexibility of the power grid [33]. In addition, low-carbon fuels such as hydrogen, ammonia and methanol have further broadened the choice of transportation to connect with future energy ecosystems, especially in the field of commercial and heavy-duty transportation.

(1)

Key Technology Directions for Mobility Electrification

From a system perspective, the decarbonization benefit of EV deployment mainly depends on two factors: the efficiency of vehicles converting electrical energy into power, and the flexibility of electric vehicles interacting with the power grid. These requirements go beyond traditional vehicle performance indicators and place more emphasis on grid compatibility, full-life cycle sustainability and all-weather operation stability.

Therefore, this section focuses on four vehicle side technology directions that support transportation electrification and the energy transition: high energy density batteries, high-efficiency electric drive systems, wide-bandgap power electronic devices, and intelligent thermal management. These technologies together determine the range and efficiency of the vehicle, as well as the feasibility of large-scale high-power charging and two-way energy interaction.

(a)

High-Energy-Density Batteries: Solid-State Batteries

Battery technology remains a primary determinant of EV range, cost, safety, and lifecycle sustainability. As shown in Figure 14, the roadmap for power battery development sets increasingly aggressive targets for next-generation systems, with solid-state battery planning indicating gravimetric energy density goals of >400 Wh/kg by 2025, >500 Wh/kg by 2030, and >700 Wh/kg by 2035, together with cycle-life expectations of >1000–1500 cycles. Such targets imply that incremental improvements to conventional liquid-electrolyte lithium-ion batteries [34] may become insufficient, and that achieving both ultra-high specific energy and long cycle life will require breakthroughs in cell chemistry, materials integration, and manufacturing maturity.

In this context, all-solid-state batteries (ASSBs), particularly those coupled with lithium-metal anodes, are widely regarded as a promising pathway to improving energy density while improving safety by removing flammable liquid electrolytes. However, ASSBs are not yet automotive-ready, and their practical deployment is constrained by several engineering bottlenecks. Some studies indicate key barriers include limited ionic conductivity in some solid electrolytes, insufficient solid–solid interfacial contact, and high interfacial resistance caused by chemical and electrochemical incompatibility between electrodes and solid electrolytes. These interface-related limitations can severely limit rate capability and cycling stability, making interfacial stability and resistance among the most critical challenges to be addressed to translate solid-state lithium-metal batteries into EV-grade products [35]. From a commercialization perspective, solid-state batteries are still moving from pilot-scale validation toward automotive-scale deployment, and their large-scale adoption depends on scalable manufacturing, stable solid–solid interfaces, and effective lithium-dendrite suppression. In addition to cell chemistry, recent reviews on automotive battery management systems emphasize that future battery development also depends on advanced BMS functions, including accurate state estimation, safety monitoring, thermal management, and lifecycle optimization. Therefore, high-energy-density batteries should be developed together with intelligent BMS technologies to ensure safety, reliability, and long-term performance in EV applications [36].

(b)

High Efficiency Electric Drive Systems

The electric drive system, which consists of the traction motor, inverter, and mechanical transmission, directly affects the energy consumption, drive performance and system efficiency of the vehicle. In the process of continuous energy revolution and transformation into large-scale electrification, the next-generation electric drive system must achieve high load cycle efficiency and high power density at the same time to reduce on-board energy loss and extend the range. Integrated electric axle designs combine the motor, inverter, and gearbox into a compact unit, reducing packaging volume and parasitic losses. However, this integration also introduces new challenges in thermal management and electromagnetic interference.

At the component level, permanent magnet synchronous motors (PMSMs) still dominate because of their high torque density and high efficiency. In addition, induction motors and switched reluctance motors have attracted increasing attention because of their robustness and cost advantages. Specifically, reducing loss and optimizing control strategies are important to maximize mechanical efficiency and power-to-weight ratio. In power electronics, the improvement of topology and semiconductor devices (especially wide-bandgap devices) is important to reduce inverter loss and improve operating efficiency. Recent research indicates that high drive-cycle efficiency and high power density are difficult to achieve through isolated optimization of the motor, inverter, or controller alone [37]. Instead, coordinated design is required because motor loss, inverter switching loss, cooling capability, and control strategy are strongly coupled under real driving cycles. For example, a higher inverter switching frequency can improve the current-control bandwidth and reduce torque ripple, but it may also increase switching loss and thermal stress unless WBG devices and cooling structures are designed accordingly.

(c)

Third-Generation Wide-Bandgap Power Devices: Enabling High-Efficiency Conversion and Bidirectionality

Third-generation WBG power semiconductor materials, such as SiC and GaN, have become important candidates for next-generation EV power electronics. Compared with silicon devices, SiC and GaN offer higher breakdown electric field, faster switching capability, and better high-temperature operation. These properties allow traction inverters and onboard power converters to operate at higher switching frequencies with lower power dissipation, which can reduce the size of passive components and cooling systems. Recent reviews show that WBG-based EV traction drives can improve power density and thermal performance compared with conventional Si-based systems [38].

Despite these benefits, the use of SiC and GaN in real applications still involves trade-offs among reliability, cost, and overall system design. SiC devices are therefore widely applied in high-voltage traction inverters, whereas GaN devices are mainly used in medium-power converters such as onboard chargers and DC–DC modules. The rapid switching behavior of WBG devices introduces challenges for EMI control, insulation reliability, and thermal management. These challenges should be mitigated by further optimizing device, packaging, and cooling solutions. Therefore, the practical benefits of WBG devices depend not only on the semiconductor material itself but also on packaging parasitics, gate-driver design, EMI control, thermal management, and reliability-oriented system optimization.

(d)

Intelligent Thermal Management: A System-Level Lever for All-Climate Global Performance

Thermal management has become another system-level challenge for EVs. In general, temperature can influence battery resistance, power capability, the degradation rate, and the safety margin. The energy cost of cabin conditioning should also be considered, especially in cold or hot climates, where heating, ventilation, and air conditioning (HVAC) systems can materially reduce driving range. As EV powertrains move toward greater integration and higher power density, thermal coupling among the battery, inverter/motor, and cabin becomes stronger, making isolated component-level cooling increasingly insufficient [39].

Therefore, intelligent thermal management is shifting from passive temperature regulation to integrated architecture and predictive control. For example, by coordinating heat pumps, waste heat recovery and shared cooling loops, energy consumption can be reduced while ensuring the temperature safety of parts and passengers. From this perspective, thermal management should be optimized at the vehicle level to balance efficiency, range, fast charging capacity, durability and reliability under different working conditions.

(2)

Vehicle–Energy-Grid-Integrated Energy Ecosystem Based on Low-Carbon Energy

Low-carbon mobility ecosystems must develop in parallel with transportation electrification and energy decarbonization. As internal combustion engine vehicles gradually transition to EVs and low-carbon solutions, energy consumption is shifting from fossil fuels to renewable electricity and low-carbon fuels. This shift tightly links the transportation with the power grid and renewable energy systems, creating an integrated ecosystem of vehicle–energy-grid, as shown in Figure 15.

(a)

Comparison and Development Trends of Hybrid Power Technology

Among hybrid technology routes, Hybrid Electric Vehicle (HEV) technology is mature and requires no charging, but its batteries are too small, and it relies on petroleum, classifying it as an energy-saving vehicle rather than a new energy vehicle. A Plug-in Hybrid Electric Vehicle (PHEV) can be powered by either gasoline or electric power and has a long range, but its batteries are relatively small, and the experience of driving with a low battery is poor. A Range Extender Electric Vehicle (REEV) offers a driving experience close to pure electric and has a simple structure, but its high-speed energy efficiency is low. None of these three types strike a balance between the pure electric experience, all-scenario efficiency, and the convenience of refueling.

In the next three to five years, hybrid technology will show a clear convergence trend: the role of HEV will weaken, PHEV will evolve towards range extension, REEV will upgrade towards high efficiency, and finally evolve towards the technical route of high range, high efficiency, and high voltage. Advanced hybrid power systems for the future should possess the following seven characteristics, as shown in

Table 4. Next-generation hybrid core features.

Category	Features	Key Indicators
Three Highs	High pure electric range	Actual pure electric range >200 km (sufficient for weekly commuting, one charge per week)
	High-efficiency engine and high-power generator	Dedicated hybrid engine max thermal efficiency ≥45%; P1 generator power ≥100 kW
	High voltage platform	800 V+ high voltage architecture with SiC devices, significantly improving energy conversion efficiency
Three Lows	Low energy consumption	WLTC combined energy consumption ≤0.6 L/100 km (fuel portion)
	Low vehicle operating cost	Total operating cost close to that of pure electric vehicles (including purchase, energy, and maintenance)
	Low charging time	Pure electric charging time 10–80% SOC ≤ 10 min
One Smart	Smart energy management	AI-based full-condition energy distribution, supporting predictive energy recovery and V2G interaction

, making hybrid technology a strategic route with independent value, running parallel to pure electric systems.

(b)

An Integrated Ecosystem of Electric Vehicles and Renewable Energy

In the integrated ecosystem, pure electric vehicles and fuel cell electric vehicles are not only travel tools but also channels for energy flow. In order to meet the challenges of renewable energy and the rapid growth of demand for high-power charging, it is important to integrate photovoltaic energy storage and charging/power exchange infrastructure into the future ecosystem. By deploying on-site renewable energy, fixed energy storage and charging facilities, a new ecosystem can be built: the new energy flow will replace the traditional carbon cycle path, support local energy balance and load adjustment, improve the effective use of low-carbon electricity, and relieve the pressure on the distribution grid.

At the system level, the wide application of V2G technology enables the power grid to actively participate in peak cutting, frequency regulation and renewable energy balance using electric vehicles [3]. This two-way interaction enables the transportation sector to contribute flexible capacity to system-level operation while creating additional economic value for car owners and fleet operators. Overall, the integration of renewable energy generation, energy storage, charging infrastructure and V2G has great economic and ecological potential and is expected to achieve carbon neutrality by 2060.

To quantitatively assess the low-carbon potential of different energy pathways within this integrated ecosystem, a life cycle greenhouse gas (GHG) emission analysis is conducted for typical transportation energy carriers, including fossil fuels, low-carbon fuels, zero-carbon fuels, and electric powertrains. The well-to-wheel (WTW) emission methodology is adopted to ensure a consistent comparison across all the pathways.

For all fuel and electricity pathways, the total WTW emission CO_2,WTW is calculated as follows:

$$C{O}_{2,WTW}=C{O}_{2,WTT}+C{O}_{2,TTW}$$

(4)

where CO_2,WTT is the emissions from fuel/electricity production and transportation (Well-to-Tank) and CO_2,TTW is the emissions from vehicle end-use (Tank-to-Wheel).

For battery electric vehicles, the WTW emission CO_2,WTW,EV is directly linked to the grid carbon intensity and vehicle energy consumption:

$$C{O}_{2,WTW,EV}=E{F}_{grid}\times EC$$

(5)

where EF_grid denotes the grid carbon emission factor (g CO_2eq/kWh) and EC donates the real-world vehicle energy consumption (kWh/100 km).

The emission results in

Table 5. Life cycle greenhouse gas emissions of typical transportation energy pathways (passenger car, g CO_2eq/kWh).

Energy Type	Emission Breakdown	WTW Emissions (g CO2eq/kWh)	Normalized (Gasoline = 1.00)
Gasoline (Gasoline + 7 vol.% biofuel)	WTT: 21.6 g CO2eq/MJ TTW: 70.0 g CO2eq/MJ	235	1
Diesel (Diesel + 7 vol.% biofuel)	WTT: 22.5 g CO2eq/MJ TTW: 68.1 g CO2eq/MJ	234	0.99
Natural Gas (CNG + 3.4 vol.% biomethane)	WTT: 15.9 g CO2eq/MJ TTW: 57.9 g CO2eq/MJ	203	0.86
Green Methanol	WTT: 10–15 g CO2eq/MJ TTW: 45–50 g CO2eq/MJ	90–110	0.38–0.47
Green Ammonia	WTT: 8–12 g CO2eq/MJ TTW: 0 (carbon-free)	40–55	0.17–0.23
Green Hydrogen (Fuel Cell)	WTT: 8–12 g CO2eq/MJ TTW: 0 (carbon-free);	35–50	0.15–0.21
EV (Fossil-dominated Grid)	Grid factor: 260 g CO2eq/MJ; Real-world energy consumption: 20.2 kWh/100 km	145–155	0.62–0.66
EV (Renewable Power)	Grid factor: 21 g CO2eq/MJ (wind + solar PV); Real-world energy consumption: 20.2 kWh/100 km	52	0.22

Data sources: [40,41].

align with the carbon reduction trajectory outlined in Figure 15. By 2025, a shift to natural gas and early electrification can achieve a 15% reduction in transportation emissions compared to 2020 levels. By 2030, the widespread adoption of EVs (supported by a decarbonizing grid) and green methanol can deliver a 30% reduction. By 2040, the integration of green ammonia, green hydrogen, and high-renewable-power EVs will drive a 40% reduction, laying the foundation for full carbon neutrality by 2060.

(3)

An Example of Hydrogen-based Commercial Mobility as a Complementary Pathway

Although battery-powered travel is considered the preferred option to meet future transportation needs, some companies are exploring hydrogen-based supplements. These solutions provide low-carbon solutions for specific user groups by providing long range, high utilization and short hydrogen filling time. At present, original equipment manufacturers can choose two power system schemes: one is based on fuel cells, which is more efficient and has greater potential but needs to further reduce vehicle-level cost; the other is based on hydrogen internal combustion engines, which uses a mature engine platform. Although it is less efficient, its investment cost and vehicle-level cost are lower, which can better adapt to specific needs.

In addition to power systems, hydrogen can also act as a cross-field energy carrier, connecting renewable energy generation, long-term energy storage and carbon management. Recent research on integrated hydrogen energy storage and carbon capture and utilization technology of microgrids shows that by coordinating multi-energy scheduling, surplus renewable electricity can be absorbed, energy transfer across time scales can be achieved, and energy costs and carbon emissions can be reduced [42]. This example illustrates how hydrogen commercial transportation forms a low-carbon industrial cycle, connecting renewable energy supply, hydrogen production, vehicle deployment and fleet demand so as to supplement the battery electric charging ecosystem within a unified vehicle-energy-grid framework.

4. Conclusions

This article puts forward three new basic paradigms to guide the development of automotive engineering and emphasizes that the automobile industry must be deeply integrated with the fields of physics, information, and humanities. At the same time, industrial upgrading should be optimized through the integration theory of “Four Networks and Four Flows”. By combining the bidirectional energy flow capability of electric vehicles with the autonomous learning capability of artificial intelligence, electric vehicles are evolving into intelligent terminals and becoming key nodes of intelligent power grids.

This paper analyzed the three paradigms, the 4N4F framework, and the Human–Vehicle–Road–Cloud–Satellite (HVRCS) architecture into a unified framework for future automotive engineering comprehensively. In this framework, the three paradigms define the fundamental principles of automotive evolution, 4N4F provides the system-level structure of interacting networks and flows, and HVRCS offers the engineering architecture for practical implementation. The future automobile can be understood as a super-intelligent lifeform: a mobile intelligent space supported by energy circulation, data-driven intelligence, system-level connectivity, and human-centered interaction. Since this paper is primarily theoretical, future work should validate this integration through case studies, system modeling, and engineering demonstrations while also developing measurable indicators for interactions among the four networks and four flows. The proposed framework provides a scientific blueprint for the sustainable, intelligent, and human-centered development of the future automobile industry. Moreover, the development of automobiles must be guided by policies and regulations in the humanistic field and use information flow to drive the flow of energy and material so as to realize the flow of value and ultimately create huge economic value and social welfare.