Reflective and self-regulating architectures in embodied and agentic AI systems

Recent advances in artificial intelligence have enabled the development of embodied and agentic systems capable of operating in dynamic, real-world environments. While these systems exhibit increasing autonomy and performance, their underlying architectures remain largely focused on optimization, control, and task execution. Consequently, critical limitations persist, including limited awareness of uncertainty, the absence of robust self-monitoring mechanisms, insufficient decision traceability, and inadequate alignment with human values in high-stakes contexts.This Research Topic aims to explore an emerging direction in AI system design: the development of reflective and self-regulating architectures. These architectures incorporate internal processes that allow systems to monitor and evaluate their own behavior, reason about uncertainty, maintain continuity of internal state, and operate within structured, human-aligned constraints.Such capabilities are particularly important for embodied and agentic systems, where decisions unfold over time, interact with the physical world, and may lead to irreversible consequences. In these settings, reliance on external control mechanisms or post hoc governance alone is often insufficient. This underscores the need for AI systems that embed intrinsic capacities for reflection, constraint awareness, and adaptive self-regulation.This Research Topic welcomes contributions from theoretical, computational, and applied perspectives, including, but not limited to:• Reflective architectures and meta-cognitive layers in AI systems • Self-monitoring and uncertainty-aware models • Persistent memory, identity, and decision traceability in autonomous agents • Human-aligned constraint systems and value-aware AI design • Distributed and multi-agent systems with internal coordination and regulation mechanisms • Embodied cognition and grounded decision-making • Safety, reliability, and governance in autonomous AI systems • Hybrid human–AI systems with structured oversight and escalation mechanisms Topic editors: CHRISTOS MORIDIS, Stéphane Lallée Source: https://www.frontiersin.org/research-topics/79675/reflective-and-self-regulating-architectures-in-embodied-and-agentic-ai-systems