Data-driven neurorobotics: multimodal learning, adaptive intelligence, and reproducible human–robot interaction

Neurorobotics is entering a new phase in which progress depends not only on advances in robotic embodiment and brain-inspired control, but also on learning from complex, heterogeneous, and multimodal data. As the field evolves, its key challenges are increasingly data-centric, including representation learning, multimodal integration, reproducible evaluation, and the development of adaptive, trustworthy, human-centered intelligent systems. Recent developments indicate that neurorobotics is no longer defined solely by physical platforms or control architectures, but is increasingly shaped by multimodal learning, interaction modeling, decision-making, and computational frameworks that generalize across tasks and environments. Research highlights growing interest in multimodal fusion, human–robot interaction, learning-based autonomy, and large-scale model integration, underscoring that data representation and model adaptability are now central scientific concerns. However, the field still faces persistent challenges, including fragmented datasets, limited benchmarking standards, inconsistent evaluation practices, and reduced comparability across studies, all of which hinder reproducibility and real-world reliability. In this context, a data-scientific perspective is timely, enabling neurorobotics to be examined through representation learning, evaluation, reproducibility, and adaptive intelligence.This collection aims to advance neurorobotics by emphasizing the central role of data, learning, and methodological rigor in the development of intelligent autonomous systems. Rather than focusing primarily on hardware, actuation, or device-level innovation, it highlights how multimodal data can be transformed into robust representations, adaptive learning strategies, interpretable decisions, and meaningful human-centered interaction.We welcome contributions on multimodal representation learning, continual and transfer learning, benchmark and dataset development, reproducible evaluation, human–robot interaction analytics, explainable intelligence, privacy-aware frameworks, and collaborative learning approaches. We also encourage studies that enhance the comparability and reliability of neurorobotics research through open resources, standardized protocols, and software-oriented methodologies. Overall, this Research Topic seeks to support a more analytically grounded neurorobotics framework in which brain-inspired intelligence is understood not only through embodiment, but also through data quality, model transparency, and reproducible behavior across tasks and contexts. We particularly welcome original research, reviews, mini reviews, methods, perspectives, brief research reports, data reports, systematic reviews, and technology and code articles on topics including, but not limited to:• Multimodal learning and cross-modal representation in neurorobotics• Brain-inspired algorithms for perception, adaptation, and decision-making• Human–robot interaction analytics and collaborative intelligent systems• Interpretable, explainable, and trustworthy models for autonomous behavior• Continual learning, transfer learning, few-shot learning, and data-efficient adaptation• Benchmark datasets, annotation strategies, and reproducible evaluation protocols• Federated, distributed, or privacy-aware learning for intelligent robotic systems• Software tools, open resources, and computational pipelines supporting neurorobotics research• Large-scale or foundation-model-inspired approaches relevant to neurorobotic intelligenceSubmissions should clearly demonstrate their relevance to neurorobotics and articulate how data-driven or learning-based methods contribute to adaptive intelligence, autonomous behavior, or meaningful human–robot interaction. Topic editors: Huyong Yan, Jie Yang, Di Wu, Zhong Chen Source: https://www.frontiersin.org/research-topics/80930/data-driven-neurorobotics-multimodal-learning-adaptive-intelligence-and-reproducible-human-robot-interaction