We are delighted to present the inaugural  “Horizons in Neurorobotics” article collection. This collection showcases high-impact, authoritative and reader-friendly review articles covering the most topical research at the forefront of Neurorobotics. All contributing authors were individually nominated by the Chief Editors of the Journal in recognition of their prominence and influence in their respective fields. The cutting-edge work presented in this article collection highlights the diversity of research performed across the entire breadth of the Neurorobotics field and reflects on the latest advances in the science and technology of embodied autonomous systems as well as their applications. Contributions to the collection are by invitation only. New articles will be added to this collection as they are published.

Chemnitz University of Technology

Université Paris-Saclay

University of Southern Denmark

Frontiers in Neurorobotics

Horizons in Neurorobotics 2022

We are pleased to present the “Horizons in neurorobotics 2022” Research Topic of Frontiers in Neurorobotics. This Research Topic of articles represents a significant contribution to the field of neurorobotics, reflecting both the depth and breadth of research currently being undertaken in this area. The articles selected for publication in this Research Topic cover a wide range of topics, from fundamental theoretical developments to practical applications, each contributing uniquely to the advancement of neurorobotics.This Research Topic features five articles from leading researchers, each of whom brings their unique perspective and expertise to the neurorobotics community. Below are brief summaries of the articles featured in this Research Topic:• When neuro-robots go wrong: a review by <a href="https://doi.org/10.3389/fnbot.2023.1112839">Khan and Olds</a> discusses the challenges and future directions of explainable AI within neuro-robotics, emphasizing the need for increased transparency in AI decisions within robotic systems. Robotic design is becoming increasingly influenced by biological brains, leading to the development of neurorobots. At the same time, there is a growing societal expectation for thorough explanations. In this article, the authors discuss the convergence of these two strands and their potential influence on the course of robotic design.• Inverse kinematics solution of 6-DOF manipulator based on multi-objective full-parameter optimization PSO algorithm by <a href="https://doi.org/10.3389/fnbot.2022.791796">Luo et al.</a> introduces a novel Particle Swarm Optimization (PSO) algorithm for robotic manipulators through inverse kinematics solutions. This multi-objective, full-parameter optimization strategy notably enhances the precision and efficiency of the manipulator's movements. It incorporates multiple optimization objectives, such as minimizing position, posture, and joint angles, ensuring a comprehensive evaluation of possible solutions. Additionally, the study refines classical PSO features, including inertia weight, learning factors, and initialization strategies, which allow for more focused and effective navigation of the solution space to optimize the search process. Comparative testing against standard benchmarks and various modern algorithms demonstrates the proposed algorithm's superior performance, showing not only heightened accuracy in resolving inverse kinematics but also substantial gains in computational efficiency. Overall, this study contributes significantly to the fields of robotic motion control and optimization algorithms.• Deep causal learning for robotic intelligence by <a href="https://doi.org/10.3389/fnbot.2023.1128591">Li</a> proposes new architectures that integrate deep learning with causal inference (deep causal learning) to improve decision-making processes in robotic systems. The study envisions that deep causal learning offers a promising solution for overcoming challenges in real-world intelligent robots. This learning approach helps uncover causal relationships, enhancing our understanding of how data are generated. Such understanding boosts safety, reliability, and addresses ethical and societal concerns. Deep causal learning can be used to identify the key factors affecting a specific outcome, thereby simplifying domain-specific knowledge. It can be used to generate counterfactual predictions, enhancing decision-making and enabling complementary perception and understanding. Importantly, it organizes knowledge structures, facilitating stackable learning for improved perception, control, and scalability. The study also discusses causal cognition and statistical causal learning. Overall, it highlights a learning approach toward real-world robotic intelligence.• Continuous semi-autonomous prosthesis control using a depth sensor on the hand by <a href="https://doi.org/10.3389/fnbot.2022.814973">Castro and Dosen</a> presents a new control system for myoelectric prostheses to improve the usability and dexterity for amputees. While conventional control approaches require the placement of surface electromyography (sEMG) and/or camera sensors on the user, this study employs a depth sensor placed on the dorsal side of the prosthetic hand, facilitating the semi-autonomous control of grasping actions. This practical approach eliminates the need for the user to wear additional equipment. The experimental results illustrate that the participants could use the system to grasp a variety of objects or object parts, approached from different directions, and placed either individually or within a cluttered environment. The system allows the participants to easily learn to grasp an object. Overall, the study represents an important step toward the development of a self-contained semi-autonomous system and reflects the advancement in the science and technology of embodied neurorobotics in clinical applications.• Recent advancements in multimodal human–robot interaction by <a href="https://doi.org/10.3389/fnbot.2023.1084000">Su et al.</a> reviews advances in human-robot interaction, focusing on multimodal approaches that enhance interactions between humans and robots. The authors thoroughly analyze the progress and upcoming trends in multi-modal HRI by examining possible I/O signals, modalities, and feedback mechanisms. Almost 100 research articles from all major application areas are summarized.The field of neurorobotics, which originally focused on embodied autonomous systems driven by brain-inspired mechanisms/models and/or actual biological systems, is evolving at an unprecedented pace, with new technologies and methodologies continually emerging. Looking forward, we are excited about the potential of these developments to further revolutionize our understanding and capabilities in this field. Given its multidisciplinary nature, integrating neuroscience, robotics, and artificial intelligence, the field can serve as a foundation for comprehending natural intelligence and translating it into nature-inspired machine intelligence.The articles in this Research Topic collectively push the boundaries of neurorobotics, offering new insights and tools that will undoubtedly influence future research directions. We hope that this Research Topic will serve as a valuable resource for both newcomers to the field and established researchers.FR: Writing &#x2013; original draft, Writing &#x2013; review &amp; editing. HS: Writing &#x2013; original draft, Writing &#x2013; review &amp; editing. PM: Writing &#x2013; original draft, Writing &#x2013; review &amp; editing.The author(s) declare that no financial support was received for the research, authorship, and/or publication of this article.The success of this Research Topic owes much to the diligent efforts of our authors, reviewers, and the editorial board. We thank them for their commitment to excellence and their contributions that have made this publication possible.The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.The author(s) declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision.All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Editorial: Horizons in neurorobotics 2022

Robotics have advanced significantly over the years, and human–robot interaction (HRI) is now playing an important role in delivering the best user experience, cutting down on laborious tasks, and raising public acceptance of robots. New HRI approaches are necessary to promote the evolution of robots, with a more natural and flexible interaction manner clearly the most crucial. As a newly emerging approach to HRI, multimodal HRI is a method for individuals to communicate with a robot using various modalities, including voice, image, text, eye movement, and touch, as well as bio-signals like EEG and ECG. It is a broad field closely related to cognitive science, ergonomics, multimedia technology, and virtual reality, with numerous applications springing up each year. However, little research has been done to summarize the current development and future trend of HRI. To this end, this paper systematically reviews the state of the art of multimodal HRI on its applications by summing up the latest research articles relevant to this field. Moreover, the research development in terms of the input signal and the output signal is also covered in this manuscript.

Robotics have advanced significantly over the years, and human&#x2013;robot interaction (HRI) is now playing an important role in delivering the best user experience, cutting down on laborious tasks, and raising public acceptance of robots. New HRI approaches are necessary to promote the evolution of robots, with a more natural and flexible interaction manner clearly the most crucial. As a newly emerging approach to HRI, multimodal HRI is a method for individuals to communicate with a robot using various modalities, including voice, image, text, eye movement, and touch, as well as bio-signals like EEG and ECG. It is a broad field closely related to cognitive science, ergonomics, multimedia technology, and virtual reality, with numerous applications springing up each year. However, little research has been done to summarize the current development and future trend of HRI. To this end, this paper systematically reviews the state of the art of multimodal HRI on its applications by summing up the latest research articles relevant to this field. Moreover, the research development in terms of the input signal and the output signal is also covered in this manuscript.

Recent advancements in multimodal human–robot interaction

This invited Review discusses causal learning in the context of robotic intelligence. The Review introduces the psychological findings on causal learning in human cognition, as well as the traditional statistical solutions for causal discovery and causal inference. Additionally, we examine recent deep causal learning algorithms, with a focus on their architectures and the benefits of using deep nets, and discuss the gap between deep causal learning and the needs of robotic intelligence.

This invited Review discusses causal learning in the context of robotic intelligence. The Review introduces the psychological findings on causal learning in human cognition, as well as the traditional statistical solutions for causal discovery and causal inference. Additionally, we examine recent deep causal learning algorithms, with a focus on their architectures and the benefits of using deep nets, and discuss the gap between deep causal learning and the needs of robotic intelligence.

Deep causal learning for robotic intelligence

Neuro-robots are a class of autonomous machines that, in their architecture, mimic aspects of the human brain and cognition. As such, they represent unique artifacts created by humans based on human understanding of healthy human brains. European Union’s Convention on Roboethics 2025 states that the design of all robots (including neuro-robots) must include provisions for the complete traceability of the robots’ actions, analogous to an aircraft’s flight data recorder. At the same time, one can anticipate rising instances of neuro-robotic failure, as they operate on imperfect data in real environments, and the underlying AI behind such neuro-robots has yet to achieve explainability. This paper reviews the trajectory of the technology used in neuro-robots and accompanying failures. The failures demand an explanation. While drawing on existing explainable AI research, we argue explainability in AI limits the same in neuro-robots. In order to make robots more explainable, we suggest potential pathways for future research.

Neuro-robots are a class of autonomous machines that, in their architecture, mimic aspects of the human brain and cognition. As such, they represent unique artifacts created by humans based on human understanding of healthy human brains. European Union&#x2019;s Convention on Roboethics 2025 states that the design of all robots (including neuro-robots) must include provisions for the complete traceability of the robots&#x2019; actions, analogous to an aircraft&#x2019;s flight data recorder. At the same time, one can anticipate rising instances of neuro-robotic failure, as they operate on imperfect data in real environments, and the underlying AI behind such neuro-robots has yet to achieve explainability. This paper reviews the trajectory of the technology used in neuro-robots and accompanying failures. The failures demand an explanation. While drawing on existing explainable AI research, we argue explainability in AI limits the same in neuro-robots. In order to make robots more explainable, we suggest potential pathways for future research.

When neuro-robots go wrong: A review

Modern myoelectric prostheses can perform multiple functions (e.g., several grasp types and wrist rotation) but their intuitive control by the user is still an open challenge. It has been recently demonstrated that semi-autonomous control can allow the subjects to operate complex prostheses effectively; however, this approach often requires placing sensors on the user. The present study proposes a system for semi-autonomous control of a myoelectric prosthesis that requires a single depth sensor placed on the dorsal side of the hand. The system automatically pre-shapes the hand (grasp type, size, and wrist rotation) and allows the user to grasp objects of different shapes, sizes and orientations, placed individually or within cluttered scenes. The system “reacts” to the side from which the object is approached, and enables the user to target not only the whole object but also an object part. Another unique aspect of the system is that it relies on online interaction between the user and the prosthesis; the system reacts continuously on the targets that are in its focus, while the user interprets the movement of the prosthesis to adjust aiming. Experimental assessment was conducted in ten able-bodied participants to evaluate the feasibility and the impact of training on prosthesis-user interaction. The subjects used the system to grasp a set of objects individually (Phase I) and in cluttered scenarios (Phase II), while the time to accomplish the task (TAT) was used as the performance metric. In both phases, the TAT improved significantly across blocks. Some targets (objects and/or their parts) were more challenging, requiring thus significantly more time to handle, but all objects and scenes were successfully accomplished by all subjects. The assessment therefore demonstrated that the system is indeed robust and effective, and that the subjects could successfully learn how to aim with the system after a brief training. This is an important step toward the development of a self-contained semi-autonomous system convenient for clinical applications.

Modern myoelectric prostheses can perform multiple functions (e.g., several grasp types and wrist rotation) but their intuitive control by the user is still an open challenge. It has been recently demonstrated that semi-autonomous control can allow the subjects to operate complex prostheses effectively; however, this approach often requires placing sensors on the user. The present study proposes a system for semi-autonomous control of a myoelectric prosthesis that requires a single depth sensor placed on the dorsal side of the hand. The system automatically pre-shapes the hand (grasp type, size, and wrist rotation) and allows the user to grasp objects of different shapes, sizes and orientations, placed individually or within cluttered scenes. The system &#x201c;reacts&#x201d; to the side from which the object is approached, and enables the user to target not only the whole object but also an object part. Another unique aspect of the system is that it relies on online interaction between the user and the prosthesis; the system reacts continuously on the targets that are in its focus, while the user interprets the movement of the prosthesis to adjust aiming. Experimental assessment was conducted in ten able-bodied participants to evaluate the feasibility and the impact of training on prosthesis-user interaction. The subjects used the system to grasp a set of objects individually (Phase I) and in cluttered scenarios (Phase II), while the time to accomplish the task (TAT) was used as the performance metric. In both phases, the TAT improved significantly across blocks. Some targets (objects and/or their parts) were more challenging, requiring thus significantly more time to handle, but all objects and scenes were successfully accomplished by all subjects. The assessment therefore demonstrated that the system is indeed robust and effective, and that the subjects could successfully learn how to aim with the system after a brief training. This is an important step toward the development of a self-contained semi-autonomous system convenient for clinical applications.

Continuous Semi-autonomous Prosthesis Control Using a Depth Sensor on the Hand

A multi-objective full-parameter optimization particle swarm optimization (MOFOPSO) algorithm is proposed in this paper to overcome the drawbacks of poor accuracy, low efficiency, and instability of the existing algorithms in the inverse kinematics(IK) solution of the manipulator. In designing the multi-objective function, the proposed algorithm considers the factors such as position, posture, and joint. To improve PSO, the proposed algorithm comprehensively analyzes all factors affecting the global and local searching abilities. Based on this, the initial population is designed following the localized uniform distribution method. Meanwhile, the inertia weight, asynchronous learning factor, and time factor are respectively designed by introducing the iteration factor. Finally, this paper tests the performance of MOFOPSO with three typical functions to obtain a better inverse kinematics solution of the 6-DOF manipulator. Also, six other algorithms are taken for performance comparison. The experimental results indicate that the proposed method not only ensures the stability of the manipulator but also achieves high accuracy and efficiency in solving the inverse kinematics of the 6-DOF manipulator.

A multi-objective full-parameter optimization particle swarm optimization (MOFOPSO) algorithm is proposed in this paper to overcome the drawbacks of poor accuracy, low efficiency, and instability of the existing algorithms in the inverse kinematics(IK) solution of the manipulator. In designing the multi-objective function, the proposed algorithm considers the factors such as position, posture, and joint. To improve PSO, the proposed algorithm comprehensively analyzes all factors affecting the global and local searching abilities. Based on this, the initial population is designed following the localized uniform distribution method. Meanwhile, the inertia weight, asynchronous learning factor, and time factor are respectively designed by introducing the iteration factor. Finally, this paper tests the performance of MOFOPSO with three typical functions to obtain a better inverse kinematics solution of the 6-DOF manipulator. Also, six other algorithms are taken for performance comparison. The experimental results indicate that the proposed method not only ensures the stability of the manipulator but also achieves high accuracy and efficiency in solving the inverse kinematics of the 6-DOF manipulator.