Explore the vast range of titles available.

This paper delves into the potential of Large Language Model (LLM) agents for industrial robotics, with an emphasis on autonomous design, decision-making, and task execution within manufacturing contexts. We propose a comprehensive framework that includes three core components: (1) matches manufacturing tasks with process parameters, emphasizing the challenges in LLM agents’ understanding of human-imposed constraints; (2) autonomously designs tool paths, highlighting the LLM agents’ proficiency in planar tasks and challenges in 3D spatial tasks; and (3) integrates embodied intelligence within industrial robotics simulations, showcasing the adaptability of LLM agents like GPT-4. Our experimental results underscore the distinctive performance of the GPT-4 agent, especially in Component 3, where it is outstanding in task planning and achieved a success rate of 81.88% across 10 samples in task completion. In conclusion, our study accentuates the transformative potential of LLM agents in industrial robotics and suggests specific avenues, such as visual semantic control and real-time feedback loops, for their enhancement.

[This corrects the article DOI: 10.3389/fnbot.2026.1796043.].

The exploration of embodied intelligence has garnered widespread consensus in the field of artificial intelligence (AI), aiming to achieve artificial general intelligence (AGI). Classical AI models, which rely on labeled data for learning, struggle to adapt to dynamic, unstructured environments due to their offline learning paradigms. Conversely, embodied intelligence emphasizes interactive learning, acquiring richer information through environmental interactions for training, thereby enabling autonomous learning and action. Early embodied tasks primarily centered on navigation. With the surge in popularity of large language models (LLMs), the focus shifted to integrating LLMs/multimodal large models (MLM) with robots, empowering them to tackle more intricate tasks through reasoning and planning, leveraging the prior knowledge imparted by LLM/MLM. This work reviews initial embodied tasks and corresponding research, categorizes various current embodied intelligence schemes deployed in robotics within the context of LLM/MLM, summarizes the perception–planning–action (PPA) paradigm, evaluates the performance of MLM across different schemes, and offers insights for future development directions in this domain.

Artificial intelligence (AI), particularly through advanced large language model (LLM) technologies, is reshaping coal mine safety assessment methods with its powerful cognitive capabilities. Given the dynamic, multi-source, and heterogeneous characteristics of data in typical mining scenarios, traditional manual assessment methods are limited in their information processing capacity and cost-effectiveness. This study addresses these challenges by proposing an embodied intelligent system for mine safety assessment based on multi-level large language models (LLMs) for multi-source sensor data. The system employs a multi-layer architecture implemented through multiple LLMs, enabling not only rapid and effective processing of multi-source sensor data but also enhanced environmental perception through physical interactions. By leveraging the tool invocation and reasoning capabilities of LLM in conjunction with a coal mine safety knowledge base, the system achieves logical inference, anomalous data detection, and potential safety risk prediction. Furthermore, its memory functionality ensures the learning and utilization of historical experiences, providing a solid foundation for continuous assessment processes. This study established a comprehensive experimental framework integrating numerical simulation, scenario simulation, and real-world testing to evaluate the system through embodied intelligence. Experimental results demonstrate that the system effectively processes sensor data and exhibits rapid, efficient safety assessment capabilities during embodied interactions, offering an innovative solution for coal mine safety.

Since the emergence of ChatGPT, research on large language models (LLMs) has actively progressed across various fields. LLMs, pre-trained on vast text datasets, have exhibited exceptional abilities in understanding natural language and planning tasks. These abilities of LLMs are promising in robotics. In general, traditional supervised learning-based robot intelligence systems have a significant lack of adaptability to dynamically changing environments. However, LLMs help a robot intelligence system to improve its generalization ability in dynamic and complex real-world environments. Indeed, findings from ongoing robotics studies indicate that LLMs can significantly improve robots’ behavior planning and execution capabilities. Additionally, vision-language models (VLMs), trained on extensive visual and linguistic data for the vision question answering (VQA) problem, excel at integrating computer vision with natural language processing. VLMs can comprehend visual contexts and execute actions through natural language. They also provide descriptions of scenes in natural language. Several studies have explored the enhancement of robot intelligence using multimodal data, including object recognition and description by VLMs, along with the execution of language-driven commands integrated with visual information. This review paper thoroughly investigates how foundation models such as LLMs and VLMs have been employed to boost robot intelligence. For clarity, the research areas are categorized into five topics: reward design in reinforcement learning, low-level control, high-level planning, manipulation, and scene understanding. This review also summarizes studies that show how foundation models, such as the Eureka model for automating reward function design in reinforcement learning, RT-2 for integrating visual data, language, and robot actions in vision-language-action models, and AutoRT for generating feasible tasks and executing robot behavior policies via LLMs, have improved robot intelligence.

The global aging trend is becoming increasingly severe, and the demand for life assistance and medical rehabilitation for frail and disabled elderly people is growing. As the best solution for assisting limb movement, guiding limb rehabilitation, and enhancing limb strength, exoskeleton robots are becoming the focus of attention from all walks of life. This paper reviews the progress of research on upper limb exoskeleton robots, sEMG technology, and intention recognition technology. It analyzes the literature using keyword clustering analysis and comprehensively discusses the application of sEMG technology, deep learning methods, and machine learning methods in the process of human movement intention recognition by exoskeleton robots. It is proposed that the focus of current research is to find algorithms with strong adaptability and high classification accuracy. Finally, traditional machine learning and deep learning algorithms are discussed, and future research directions are proposed, such as using a deep learning algorithm based on multi-information fusion to fuse EEG signals, electromyographic signals, and basic reference signals. A model with stronger generalization ability is obtained after training, thereby improving the accuracy of human movement intention recognition based on sEMG technology, which provides important support for the realization of human–machine fusion-embodied intelligence of exoskeleton robots.

Engineers frequently aim to streamline environmental factors to facilitate the effective operation of robots. However, in nature, environmental considerations play a crucial role in shaping the embodiment of organisms. To comply robots with the complexity of real‐world environments, embedding similar intelligence is key. In the field of soft robotics, various approaches offer insight into how intelligence can be integrated into artificial agents. A discussed topic is the intricate relationship between the brain and the body at the core of intelligence in robots. The goal of this article is, therefore, to unravel the strategies to implement different types of intelligence currently adopted in soft robots. A classification is made by making a distinction between agents that adapt to their environment by 1) their adaptive shape, 2) their adaptive functionality, and 3) their adaptive mechanics. Additionally, the perspectives on intelligence based on their computational approach are distinguished: centralized computation, decentralized computation, or embedded computation. It is concluded that a tailored robotic design approach attuned to specific environmental demands is needed. To unlock the full potential of soft robots, a fresh perspective on embodied intelligence is described, so‐called mechanical intelligence, emphasizing the robot's responsiveness to changing external conditions of a real‐world environment. This study explores strategies for embedding intelligence in soft robots to adapt to complex real‐world environments. It classifies robots that embed intelligence by adaptive shape, functionality, and mechanics and by computational approach: centralized, decentralized, or embedded. The study concludes that a tailored design and a new perspective on embodied intelligence, called “mechanical intelligence,” is essential for optimizing soft robots.

Soft robotics, distinguished by intrinsic compliance, biomimetic adaptability, and safe human‐environment interaction, has emerged as a transformative paradigm in next‐generation intelligent systems. Biological systems, refined through evolutionary optimization, exhibit unparalleled multifunctionality in unstructured environments, inspiring the development of soft robots with energy‐efficient reconfiguration and environmental responsiveness. This review presents a comprehensive analysis of intelligent soft robotics via multidisciplinary integration, covering key aspects from bioinspired design principles to advanced functional implementation. Recent breakthroughs across four interconnected domains are systematically examined: 1) biomimetic actuation mechanisms that enhance actuation efficiency through innovative structural configurations; 2) programmable materials enabling adaptive morphology and tunable mechanical properties; 3) multiscale manufacturing techniques for fabricating complex heterogeneous structures; and 4) closed‐loop control strategies integrating artificial intelligence algorithms. While highlighting emerging applications in biomedical engineering, environmental exploration, and human‐machine interfaces, challenges such as actuation efficiency, material degradation, manufacturing limitations, nonlinear‐control complexity, and sensing instability under real‐world conditions are discussed. Furthermore, strategic research directions are identified to guide the development of next‐generation soft robots endowed with embodied intelligence and adaptive functionalities. Notably, by synergizing advances in materials science, mechanical engineering, and computational intelligence, soft robotics is poised to redefine the boundaries of intelligent machines across healthcare, exploration, and human augmentation. Soft robotics, featuring intrinsic compliance and biomimetic adaptability, emerges as transformative in next‐generation intelligent systems. This review outlines how advancements in four foundational domains—actuation, materials, manufacturing, and control—drive the evolution of bioinspired intelligent soft robotics, poised to redefine the boundaries of intelligent machines across healthcare, exploration, and human augmentation.

The synergy between material-guided cell organization and stimulus-responsive actuation is essential for developing mechanically dynamic tissue-like biomaterial constructs.Responsive biomaterials should enable durable reversible actuation under cyclical deformation, while offering design freedom over material architecture, cell distribution, actuated configurations, and actuation timescale.Integrating (bio)fabrication technologies and multi-material, multiscale design enables spatiotemporal engineering of cell microenvironments.Advanced responsive biomaterials can be designed to interact with cells in ways that foster mutual adaptation. The resulting cell–material interactions can be guided into self-reinforcing complex behaviors that resemble or surpass native tissues, with applications in tissue engineering, developmental and mechanobiology, and excitable tissue dynamics. Organized cell architecture and dynamic forces are key for (re)creating native-like tissue function (e.g., contractile soft tissues). However, few studies have explored the combined effects of material-guided 3D cell organization with mechanical stimulation. Herein we underscore the importance of converging material-driven guidance of cell organization with stimulus-responsive actuation for multiscale biomaterial design, outlining strategies to engineer such biomaterials. Given the state-of-the-art biomaterials for multiscale spatiotemporally controlled organization and actuation, we propose a synergistic approach (‘meta-adaptive biomaterials’) that unlocks complexity in engineered biomaterials, harnessing adaptive feedback pathways arising from cell–material interactions. These can be designed similarly to cell–extracellular matrix (ECM) interactions to reinforce user-specified behaviors and yield functionalities that resemble or surpass native tissues, expanding possibilities in tissue engineering, in vitro models, and biohybrid robotics. Organized cell architecture and dynamic forces are key for (re)creating native-like tissue function (e.g., contractile soft tissues). However, few studies have explored the combined effects of material-guided 3D cell organization with mechanical stimulation. Herein we underscore the importance of converging material-driven guidance of cell organization with stimulus-responsive actuation for multiscale biomaterial design, outlining strategies to engineer such biomaterials. Given the state-of-the-art biomaterials for multiscale spatiotemporally controlled organization and actuation, we propose a synergistic approach (‘meta-adaptive biomaterials’) that unlocks complexity in engineered biomaterials, harnessing adaptive feedback pathways arising from cell–material interactions. These can be designed similarly to cell–extracellular matrix (ECM) interactions to reinforce user-specified behaviors and yield functionalities that resemble or surpass native tissues, expanding possibilities in tissue engineering, in vitro models, and biohybrid robotics.

The ability to autonomously navigate and explore complex 3D environments in a purposeful manner, while integrating visual perception with natural language interaction in a human-like way, represents a longstanding research objective in Artificial Intelligence (AI) and embodied cognition. Vision-Language Navigation (VLN) has evolved from geometry-driven to semantics-driven and, more recently, knowledge-driven approaches. With the introduction of Large Language Models (LLMs) and Vision-Language Models (VLMs), recent methods have achieved substantial improvements in instruction interpretation, cross-modal alignment, and reasoning-based planning. However, existing surveys primarily focus on traditional VLN settings and offer limited coverage of LLM-based VLN, particularly in relation to Sim2Real transfer and edge-oriented deployment. This paper presents a structured review of LLM-enabled VLN, covering four core components: instruction understanding, environment perception, high-level planning, and low-level control. Edge deployment and implementation requirements, datasets, and evaluation protocols are summarized, along with an analysis of task evolution from path-following to goal-oriented and demand-driven navigation. Key challenges, including reasoning complexity, spatial cognition, real-time efficiency, robustness, and Sim2Real adaptation, are examined. Future research directions, such as knowledge-enhanced navigation, multimodal integration, and world-model-based frameworks, are discussed. Overall, LLM-driven VLN is progressing toward deeper cognitive integration, supporting the development of more explainable, generalizable, and deployable embodied navigation systems.