autonomous systems

Current autonomous vehicles (AuVs) are largely confined to environmental perception and task execution, failing to meet human-centric mobility requirements—such as social interaction, contextual reasoning, goal adaptation, tool utilization, and long-horizon planning—thereby exposing a fundamental gap between technical autonomy and human-AI collaboration. This paper introduces the “Agentic Vehicle” (AgV) paradigm, marking the first conceptual shift from *autonomy* to *agency*, and proposes a cognition-communication hierarchical framework to rigorously distinguish AgVs from conventional AuVs. Methodologically, we integrate embodied intelligence, large language models, multi-agent systems, robotic control, and human–vehicle interaction. Our contributions include: (1) a system-level architectural blueprint for AgVs; and (2) a systematic identification and formal characterization of critical challenges and developmental pathways in safety robustness, real-time decision-making, ethical alignment, and collaborative governance.

This study addresses the challenge of deploying agentic AI in regulated environments, where existing approaches lack a systematic design framework that jointly accounts for autonomy and agency, often failing to balance compliance, auditability, and error correction. The work introduces the first unified model of these two dimensions, defining a two-dimensional hierarchical design space with five operational levels each. It proposes six architectural strategies—checkpoints, escalation mechanisms, multi-agent delegation, tool provisioning, tool sandboxing, and write staging—to enable flexible system configuration under real-world regulatory constraints. Validated through public-sector case studies, the framework establishes a shared terminology and actionable design guidelines, facilitating interpretable, controllable, and compliant AI deployment amid evolving model capabilities and tool fidelity.

This work addresses the absence of a unified conceptual framework for describing the autonomy of AI agents and the allocation of decision-making authority in contemporary CI/CD pipelines. It introduces the notion of “authority transfer” to systematically delineate the boundaries of agent autonomy, distinguishing between decision rights in the data plane and the control plane, and identifies governance of the control plane as a critical research direction. Through architectural abstraction, pattern identification, and governance mechanism design—supported by prototype implementation and analysis of industrial platforms—the study reveals three prevalent patterns: constrained autonomy, externally dominated governance, and delayed evaluation. These findings establish a theoretical foundation and outline a research agenda for developing safe, controllable, and highly autonomous CI/CD systems.

Current development practices for autonomous systems lack systematic empirical investigation, hindered by dynamically evolving application scenarios and domain-specific industrial constraints. Method: This paper presents the first longitudinal empirical study, conducting an anonymous, multi-industry expert survey in 2019, integrating qualitative and quantitative analysis to examine real-world industrial practices in model-based software engineering (MBSwE) and reuse for autonomous systems. Contribution/Results: The study systematically characterizes development processes, standards adoption, and verification and validation (V&V) practices; identifies critical verification bottlenecks and process adaptation barriers; and—crucially—establishes the first comparable, traceable baseline dataset of autonomous system development practices. This dataset enables rigorous longitudinal comparison and evidence-based evolution of development methods and standards.

How to rigorously define the autonomy level of AI agents while balancing innovation potential and risk mitigation? This paper proposes the first systematic, quantifiable five-level autonomy framework, explicitly treating autonomy as a design dimension orthogonal to capability and environment, and delineating control boundaries based on user roles (from operator to observer). It introduces the novel concept of an “AI Autonomy Certificate” to enable tiered governance for both single and multi-agent systems, and establishes a new evaluation paradigm centered on human–agent interaction modalities. Integrating human–agent modeling, hierarchical design principles, governance architecture, and behavioral norms, the framework yields a calibrated, verifiable, and implementable methodology for autonomy assessment. It provides a technical pathway for developing safe, controllable AI agents and delivers an auditable certification basis for regulatory oversight.

This work addresses the common challenges in autonomous drone development—such as fragmented requirements, architectural inconsistencies, and poor traceability—stemming from disjointed design processes. To bridge these gaps, the authors propose a SysML-based model-driven systems engineering framework that integrates a unified four-layer model encompassing requirements, functions, logical components, and physical/software elements. Crucially, this approach establishes, for the first time, a deep alignment between multiple SysML diagrams—including requirement, activity, and block definition diagrams—and the ROS 2 architecture, specifically its nodes, topics, services, and actions. The framework enables end-to-end traceable design, allowing early-stage allocation of requirements, precise interface specification, clear subsystem responsibility assignment, and verification planning—all prior to simulation or deployment—thereby effectively supporting typical mission scenarios such as obstacle avoidance and return-to-home operations.

This study addresses the lack of explicit behavioral specifications and validation criteria for autonomous driving systems within their Operational Design Domain (ODD). Building upon the PEGASUS six-layer model, the authors propose a comprehensive behavioral capability taxonomy encompassing 21 capabilities across three key scenarios—highway, urban, and interchange environments—structured along longitudinal and lateral control dimensions and characterized by four attributes: safety, compliance, comfort, and efficiency. The work innovatively establishes a cross-mapping between parameterized ODD definitions and behavioral specifications, thereby introducing, for the first time, a verifiable and testable behavioral specification layer. Notably, interchange scenarios are identified as a structurally distinct and underexplored domain. Leveraging a rule-driven trajectory optimization system and aligned with standards such as SAE J3016, the proposed framework enables standardized, actionable behavioral capability assessment, supports SOTIF-compliant evidence generation, and demonstrates practical efficacy as an operational specification layer in real-world deployments.

This work addresses a critical gap in current AI systems, which often conflate technical capability with operational authority, resulting in inadequate governance of authorized autonomy. The paper proposes a structured governance framework that systematically distinguishes between an AI system’s Autonomous Capability Level (ACL) and its Authorized Autonomy Level (AAL). By integrating risk exposure, action reversibility, and accountability, the framework introduces a dynamic authorization mechanism that decouples capability from permission. Validated in enterprise-grade data engineering agents, the approach enables high-capability systems to be safely constrained to lower authorization levels aligned with organizational risk tolerance. Through layered autonomy modeling and risk-aware decision protocols, the framework ensures that AI autonomy remains both effective and responsibly governed.

This work addresses the limitations of existing vision-language models in robotic control—namely, poor interpretability, weak generalization, and reliance on cloud-based computation—by proposing a fully onboard multi-agent architecture. The system deploys lightweight vision-language models (3–20B parameters) alongside vision-language-action models on an AMD Ryzen AI mini PC, enabling autonomous mobile manipulation without external support through fine-tuning and hardware-in-the-loop simulation. A novel “Megamind” coordinating agent is introduced to mitigate the challenge of context retention in long-horizon tasks faced by smaller models. The architecture’s feasibility in terms of cost, performance, and real-world transferability is validated across five industrial warehouse tasks, and the associated simulation environment is open-sourced.