What really matters for biomedical robotics

Citation

Liu, H. (2010), "What really matters for biomedical robotics", Industrial Robot, Vol. 37 No. 4. https://doi.org/10.1108/ir.2010.04937daa.002

What really matters for biomedical robotics

Article Type: Viewpoint From: Industrial Robot: An International Journal, Volume 37, Issue 4

The author Honghai LiuInstitute of Industrial Research, University of Portsmouth, Portsmouth, UK

It is evident that growing attention has been paid to biomedical robotics in recent years from multi-disciplinary perspectives including robotics, computing, clinical medicine, physiology, and psychology. It is expected that technology of biomedical robotics could provide feasible solutions for the integration of robotic techniques and medical applications with consideration of clinical requirements and physiological knowledge. Different challenging issues, confronted by researchers in biomedical robotics, need a cross-disciplinary effort involving a variety of factors which are usually not involved in conventional robotics such as social and cultural elements.

Biomedical robotics usually can be categorized into robotic studies for surgery, rehabilitation, and medical and bio applications relating to robot assistive technologies and biologically inspired solutions. There are many successful biomedical applications assisted by robots. For instance, robotics has been employed to conduct minimal invasive surgery in Imperial College London; robotic devices have been developed for upper limb rehabilitation for stroke patients using functional electrical stimulation in Southampton University. However, there is no systematic theory for robotics in biological and medical applications, most of the robotic devices developed for individual biomedical applications are on an ad hoc basis. Apart from the complex and diverse nature of biomedical applications, its state of the art might be caused by the scientists who do the work coming from so many different disciplines. One thing for sure is that theoretical and experimental challenges posed by the application of robotics and mechatronics in medicine and biology are overwhelming. It would be a long-term target to formalize mathematical theory from practical and clinical applications through experimental science and technological development.

In pursuing the perfection of biomedical robotics, which factors really matter? Is it fast-developing innovations in relevant robotics and information technologies, physiological knowledge, end-users, or something else such as quickly getting research results into clinical practice and commercial development. The answer requires a good understanding of the advantages and drawbacks of individual systems, the challenges for future technological developments, as well as its impact on end-users in terms of social and cultural contexts. One of the key factors in biomedical robotics and its application is being end-user centred since robotic devices have to interact with human users by assisting their tasks or recovering their physiological functions. Interacting with humans opens up Pandora’s box, from the perspective of finding perfect solutions for biomedical applications, it makes formalizing a mathematical framework almost impossible due to the lack of mathematical modeling of issues regarding to relevant safety, ethics, social behavior and culture. On the other hand, direct involvement of end-users in the design of this kind of user-centred robots actually substantially guarantees the success rate of ad hoc solutions to biomedical robots, since such a robot is evaluated by analyzing end-users needs and preferences and by experimental validation of prototypes through clinical trials with sample end-users. This dilemma leads to the challenge of how to construct a physiological knowledge base in a unified mathematical term by directly involving human users, and further to engineering suitable robotic devices and support the interaction via the devices. Statistics provide a mathematically sound framework for analyzing human-centred data and constructing a knowledge base, it usually, however, cannot meet the real-time requirements of robotic systems. It is evident that lots of techniques have been proposed to handle the trade-off to some extent, such as approximation and reasoning. I would like to mention that fuzzy systems, from a practical perspective, have been successfully employed to handle the trade-off of computational cost and systems performance for robotics in general, a variety of techniques have also been proposed to generate fuzzy membership functions of physiology knowledge via learning mechanisms. Will it work for biomedical robotics?

What really matters to biomedical robotics, clearly it is no more than reconsidering old-fashion robotics issues, but in human-centred environments.