Neuro-Symbolic and Multimodal Reasoning in Autonomous AI Systems

This Research Topic is presented in association with the 13th EAI International Conference on Cloud Computing (https://cloudcomp.eai-conferences.org/2026/). As cloud platforms increasingly serve as the infrastructure backbone for large-scale AI deployment, including multimodal systems and Large Language Models, the question of how these systems reason, not just perceive, becomes central. This Research Topic extends the conference’s focus on AI–cloud convergence by addressing one of the field’s most pressing open problems: building AI systems capable of systematic, verifiable reasoning over multimodal inputs.Recent advances in multimodal AI have enabled systems to perceive images, text, speech, and sensor data with high accuracy. However, perception alone does not constitute intelligence. While computer vision and large language models have achieved strong performance in recognition and generation, they often lack systematic reasoning capabilities. Models may detect objects but fail to infer causality, generate fluent explanations without logical consistency, or act autonomously without verifiable reasoning chains.The central challenge is enabling AI systems to reason over multimodal inputs, integrate structured and unstructured knowledge, and produce logically coherent, interpretable decisions. Emerging directions such as neuro-symbolic learning, structured knowledge integration, retrieval-augmented reasoning, and causal inference aim to bridge this gap, pointing toward systems that combine multimodal perception with reasoning mechanisms that are interpretable, verifiable, and robust.This Research Topic aims to advance reasoning-centric AI systems in which perception serves as input to higher-level processes such as logical inference, causal reasoning, knowledge representation, planning, and explainable decision-making. We seek contributions that move beyond standalone perception tasks toward integrated architectures that support transparent reasoning in real-world settings.We welcome contributions addressing, but not limited to: - Neuro-symbolic reasoning over multimodal inputs - Knowledge representation and structured memory for embodied agents - Causal and counterfactual reasoning in vision-language systems - Logical planning and decision-making in autonomous agents - Multimodal reasoning benchmarks and evaluation frameworks - Explainability and interpretability of reasoning chains - Trustworthy reasoning under uncertainty - Reasoning in embodied and real-world systems - Integration of perception outputs into formal reasoning modulesSubmissions focusing solely on object recognition, segmentation, or perception accuracy without a reasoning component are outside the scope of this Research Topic. We invite original research articles, theoretical frameworks, surveys, benchmark papers, and real-world case studies demonstrating reasoning-enabled AI systems. Topic editors: Hong Qing Yu, Manuella Kadar, Antonio Liotta, YingLiang Ma Source: https://www.frontiersin.org/research-topics/79840/neuro-symbolic-and-multimodal-reasoning-in-autonomous-ai-systems