Foundation Models and Neural-Symbolic AI for Robotics

Special Issue in The International Journal of Robotics Research (IJRR)

Published: by

The field of robotics has witnessed unprecedented growth in the integration of machine learning models, especially with the recent advent of foundation models, neural-symbolic AI, generative models (large language models, diffusion models, etc), and other emerging learning models. These methods have revolutionized various tasks in computer vision, natural language processing, computer graphics, etc. However, their application in the context of robotics remains underexplored. This special issue (SI) is to fill this gap and bring together leading researchers and practitioners in robotics to shed light on the latest advancements, methodologies, and best practices in this interdisciplinary domain.

This SI aims to provide a comprehensive overview of the current landscape, from the underlying theories and in-depth reviews to the forward-looking perspectives, practical implementation, and real-world challenges. By fostering a holistic understanding and promoting collaboration between experts, we aspire to accelerate the adoption of these advanced models and drive the next wave of innovations in robotic systems, including but not limited to robotic perception, cognition, planning, and control.

We accept top-quality original (unpublished) articles and review papers. Each submission undergoes peer review and follows the standard IJRR review process. Please refer to the IJRR instructions for authors for more details. Papers will be selected based on their novelty, significance, and alignment with the special section’s focus.

This call requests clear contributions to robotics, any submission that doesn’t fit into the list of topics will be desk rejected. There is no page limit for IJRR submissions. The rule is however that a paper should be as long as necessary, and no longer: conciseness is highly valued.


  • Foundation Models in Robotics
    • Challenges of integrating foundation models with robotic systems.
    • Case studies highlighting their impact on robotic applications.
  • Neural-Symbolic AI and Reasoning in Robotics
    • Exploration of the Integration of neural networks and symbolic systems such as logical, geometrical, and physical systems.
    • Real-world applications and benefits in robotics.
    • Techniques for effective neural-symbolic reasoning in complex robotic tasks.
  • Generative Models in Robotics
    • Exploring the applications and challenges of integrating state-of-the-art generative models, such as large language models, diffusion models, and other novel generative methods in robotic systems.
    • Addressing computational demands, real-time processing needs, and context comprehension hurdles when merging generative models with robotics.
  • Other Novel Neural Learning Methods in Robotics
    • Exploration of other novel neural learning techniques in robotic systems.
    • Integration barriers and effective implementation strategies.
    • Case studies showcasing the real-world benefits to robotics.
  • Robotic Perception and Cognition
    • Advanced methodologies for robotic perception using machine learning.
    • The role of cognition in enhancing robotic tasks.
    • Innovations in sensory data processing and interpretation.
  • Neural and Neuro-Symbolic Control and Planning Systems
    • Neural and neural-symbolic learning-driven control and planning mechanisms.
    • Neuro-symbolic reinforcement learning (RL) and its applications in robotics.
    • Sample complexity reduction of RL by leveraging relational knowledge.
    • Sequential decision-making with explainable policies.
    • Out-of-distribution sequential decision-making by leveraging reasoning.
    • Challenges in real-time decision-making and path planning.
  • Challenges, Limitations, and Future Perspectives
    • Addressing issues and limitations related to the scalability and robustness of foundation models and neural-symbolic AI in robotics.
    • Ethical considerations and safety concerns.
    • Showcasing the practical impact of integrating these machine learning models in diverse robotic fields such as assistive and field robotics.
    • Lessons learned and best practices from real-world implementations.
    • Predictions and trends for the future of machine learning in robotics.
    • Potential new research areas and unexplored applications.
    • The evolution of hybrid models and their long-term implications in robotics.


  • Submission Open: Feb 6, 2024
  • Submission Close: Aug 11, 2024 (No extension will be given)
  • Review and Revision: Feb to Dec 2024
  • Publication: Jan 2025

Submission Site

Guest Editors


  • Send emails directly to
  • All Guest Editors will receive your inquiries.

Sharable PDF

  • You may also download sharable PDF for this SI via This Link.

Frequently Asked Questions (FAQ)

1. Is there a hard requirement for the submission to have real robot experiments?
  • The answer would depend on the type of your work. We believe that real robot experiments can provide valuable empirical evidence and validation for a submission, however, it’s not always a strict requirement. If your research is primarily focused on theoretical aspects, simulations, or other forms of experimentation, you can explain the rationale behind your chosen approach and how it contributes to the advancement of the field. However, if real robot experiments are feasible and add significant value to your work, they can strengthen your submission by demonstrating practical applicability and validating your proposed methods in real-world scenarios.
2. Do you accept submissions containing material previously appeared in conference proceedings?
  • IJRR accepts submissions containing material previously appeared in conference proceedings. In this case, the IJRR submission should provide a substantial extension of results, methodology, analysis, conclusions and/or implications over the conference proceedings paper. An extension is considered substantial if it offers new research results, methodology, analysis, conclusions and/or implications. The mere inclusion of more details, experiments, or discussion is typically considered not substantial. The final decision on what constitutes a substantial extension will be made by the Editorial Board.

  • Details of previous submissions (including the DOI and licensing terms) must be openly disclosed in the Novelty Statement accompanying the submission to IJRR, and all necessary permissions to re-use previously published material and attribute appropriately must be obtained by authors. Failure to disclose previously submitted material does not comply with IJRR’s code of ethics and will lead to exclusion from review.

  • The manuscript submitted to IJRR must contain a statement offering an open discussion of the differences with previous conference version(s), and explicitly cite the reference(s). The conference version(s) must be uploaded as accompanying material along with the journal submission.

3. Am I allowed to submit an extended version of a conference paper which is under review?
  • It is not acceptable that manuscripts are submitted to IJRR while they are being evaluated by other archival Journals. In case of parallel submission of partly overlapping material to a non-archival conference or workshop, this should be openly disclosed at the time of IJRR submission.

  • Conferences such as ICRA, IROS, RSS, and CoRL are considered archival. Authors need to ensure any ongoing reviews at those venues are concluded before submission.

  • It is also not acceptable to submit to IJRR manuscripts which have been previously rejected anywhere else, without openly informing and discussing how the reviews received from other members of the same community have been used to improve the quality of the paper. Proper practice is to enclose all relevant materials from previous submission(s) with the IJRR submission, as supplemental files. These include information on the venue of previous submission(s), the reviews received, the answers to such reviews, and the highlights of changes in the new manuscript that address the criticisms received. This material can be prepared in a similar style as when preparing a revised version for the same Journal.

  • Manuscripts submitted elsewhere without informing the Editorial Board nor following the above practices will be editorially rejected before review. The Editorial Board of IJRR will inform the EiC and Board of other involved Journals of such decisions.

4. Are preprints allowed?
  • IJRR welcomes posting of preprint versions of an article on the author’s personal or institutional website or on community preprint servers such as ArXiv. Preprints are not regarded as prior publication. Authors should disclose details (DOI, licensing terms) of preprint posting in the Novelty Statement accompanying the submission.

  • Should authors post or update a preprint version of a manuscript that was revised after receiving feedback from the IJRR Board, it is expected that they acknowledge it in the preprint. When a manuscript is accepted and published in IJRR, it is required that the authors update the pre-print with a publication reference, including the DOI and a URL link to the published version of the article on the journal website.