An international team of experts, including a Western University electrical engineer and medical robotics expert, is recommending ways of taming the ethical, regulatory and legal frontier that would allow for a future of increasing levels of autonomy in medical robotics.

Well–defined rules for designing and operating autonomous vehicles are gaining traction around the world – but currently there’s no such framework governing the fast–growing and diverse field of medical robotics, says Western University’s Rajni Patel.

“We need to clarify what roles human and machine could play in techniques and procedures now that the technology is evolving more quickly than regulatory, ethical and legal discussions,” said Patel, professor in the Department of Electrical and Computer Engineering at Western University and Director of Engineering at Canadian Surgical Technologies and Advanced Robotics (CSTAR), Lawson Health Research Institute.

He is a co–author of a newly published editorial in the journal Science Robotics that proposes defining six levels of autonomy for medical robotic systems: ranging from fully human–controlled devices currently in wide use across the world, to the science–fiction world of robotic surgeons that “think” and are fully autonomous. The editorial is based on a session at the 2016–2017 Halcyon Dialogue on “Evolving Capabilities and Impact of Robots in Medicine” organized by the S&R Foundation and the American Association for the Advancement of Science. The meeting involved participants from computer science, engineering, medicine, and regulatory agencies with expertise and interests in various aspects of medical robotics.

At the first of six levels, the lowest, would be a zero–autonomy robot: one that follows an operator’s commands without learning to do any jobs itself. The highest level of classification would be full autonomy where human intervention would not be needed. Increasing levels of autonomy will involve developing robotic technology capable not only of performing a variety of complex tasks but also of learning from, and adapting to, changing or new situations.

Patel said technology in use today falls into the first two levels where there is no autonomy or only limited robotic assistance. Some current research in surgical robotics has developed technology capable of operating at the third level of autonomy: for example, automated suturing or automated tissue palpation that is initiated and monitored by a surgeon.

“The issues with regard to autonomy in surgical robotics are particularly complex: How do we quantify the skills that the surgical robot needs to perform autonomously and how do we ‘train’ the robot to acquire these skills? What constraints should we impose on the robots?

“As robots become more autonomous, these debates are going to become more complicated and intense. Underlying all this are the increasingly complex regulatory challenges that will need to be addressed as the level of autonomy increases.”

As the field of medical robotics grows, several well–known companies in the areas of medical devices, robotics and artificial intelligence have started working on the next generation of ever–more–advanced robotic systems. This has implications for safety, regulation, certification, training and accreditation. It also raises significant questions about cyber–security, privacy and risk of malfunction.

“We need to start asking these questions in a systematic way, long before the next generation of ‘smart’ autonomous medical robots enters every–day medical practice,” Patel said.