Call for Papers: Special Issue on Neuro-Symbolic Embodied Intelligence: Integrating Learning, Reasoning, and Acting in the Physical World

Special Issue on Neuro-Symbolic Embodied Intelligence: Integrating Learning, Reasoning, and Acting in the Physical World

Neuro-Symbolic AI combines the complementary strengths of neural and symbolic approaches: neural models excel at perception, representation learning, and generalization from data, while symbolic methods provide structured knowledge, compositional reasoning, interpretability, and mechanisms for verification. This integration is particularly relevant in embodied intelligence, where autonomous agents must interact with complex environments, interpret perceptual observations, reason about objects and their affordances, formulate goals, and execute sequences of actions under uncertainty.

Despite recent advances in foundation models, LLM-based agents, and robot learning, several fundamental challenges remain open. Purely neural approaches may produce unreliable plans, struggle with long-horizon dependencies, and lack guarantees about their behavior. Conversely, purely symbolic systems often rely on manually specified knowledge and struggle to cope with incomplete models, perceptual uncertainty, and the variability of real-world environments. Neuro-symbolic approaches offer a promising direction for addressing these limitations by integrating learned perception and control with explicit knowledge representations, logical reasoning, planning, and verification.

 

The Special Issue will address, among others, the following interconnected research directions.

  1.  Long-Horizon Planning and Self-Evolving Embodied Agents: Embodied tasks frequently require complex sequences of interdependent actions, often under partial observability and changing environmental conditions. Neuro-symbolic methods can combine symbolic planning and causal reasoning with neural perception and low-level control, enabling agents to pursue long-term goals while adapting to unexpected events.
  2. LLMs, Knowledge Reliability, and Trustworthy Embodied Agents: LLMs and foundation models provide powerful capabilities for interpreting instructions, eliciting domain knowledge, generating plans, and interacting with humans. However, their integration into embodied systems raises fundamental questions concerning reliability, consistency, grounding, and the verification of generated knowledge and actions.
  3. Semantic Representations and Knowledge Grounding for Embodied AI: Effective knowledge reuse and reasoning across different environments require semantic representations connecting objects, actions, affordances, robot capabilities, and execution semantics. Despite significant progress in robotics ontologies, knowledge graphs, and semantic mapping, these representations remain fragmented and often difficult to integrate with modern learning-based systems.

 

The goal of this Special Issue is to provide a forum for advancing the theoretical foundations, architectures, algorithms, and applications of Neuro-Symbolic Embodied Intelligence. We invite original research contributions investigating how neural and symbolic approaches can be integrated to develop intelligent embodied agents capable of learning, reasoning, planning, and acting in physical or simulated environments. We particularly encourage contributions addressing knowledge acquisition and grounding, long-horizon planning, trustworthy autonomous behavior, and the integration of foundation models with symbolic reasoning and robotic systems.

 

Topics of Interest:

Topics of interest include (but not limited to):

 

Long-Horizon Planning and Self-Evolving Embodied Agents:

• Neuro-symbolic planning and hierarchical task planning.

• Integration of symbolic planning with reinforcement learning.

• Temporal reasoning, temporal logic, and non-Markovian decision-making.

• Knowledge acquisition and symbolic model learning from interaction.

• Self-evolving agents and autonomous knowledge revision.

• Plan monitoring, failure detection, and recovery.

• Multi-agent embodied systems and coordination.

 

LLMs, Knowledge Reliability, and Trustworthy Embodied Agents:

• Integration of LLMs and foundation models into neuro-symbolic architectures.

• LLM-based planning and knowledge acquisition for embodied agents.

• Symbolic verification and constraint-based validation of generated plans.

• Grounding language and symbolic concepts in perceptual observations and actions.

• Knowledge provenance, uncertainty, and consistency management.

• Explainability, interpretability, and trustworthy autonomous behavior.

• Human-robot interaction and human-in-the-loop reasoning.

 

Semantic Representations and Knowledge Grounding for Embodied AI

• Ontologies and knowledge graphs for embodied intelligence.

• Semantic mapping and scene understanding.

• Symbol grounding and the integration of perception with structured knowledge.

• Representation of objects, actions, affordances, and robot capabilities.

• Knowledge integration and semantic interoperability across environments.

• Transferable representations for embodied reasoning.

• Neuro-symbolic knowledge engineering and automated ontology construction.

 

Deadline

31 May 2027

Papers submitted before the submission deadline will be processed according to the journal's editorial procedures.

Guest Editors

  • Elena Umili, Sapienza University of Rome, Italy
  • Emanuele Musumeci, Sapienza University of Rome, Italy
  • Vincenzo Suriani, Sapienza University of Rome, Italy
  • Daniel Dobriy, Vienna University of Economics and Business, Austria
  • Anna Sofia Lippolis, Institute of Cognitive Sciences and Technologies (ISTC-CNR), Italy

Contact email for the guest editors: nesy-embodied@googlegroups.com

Author Guidelines

We invite original research contributions addressing the topics of this Special Issue. Submissions must be original and should not have been published previously or be under consideration for publication elsewhere while being evaluated for this Special Issue. Extended versions of previously published conference or workshop papers may be considered in accordance with the journal's submission guidelines.

Submissions shall be made through the journal website:

https://neurosymbolic-ai-journal.com/

Prospective authors must follow the submission guidelines available at:

https://neurosymbolic-ai-journal.com/content/author-guidelines

Please indicate in the cover letter that the manuscript is submitted to the “Special Issue on NeSy Embodied Intelligence”.

All manuscripts will be reviewed according to the journal's open and transparent peer-review policy and will be made available online during the review process.