What are the Laws of Neurorobotics?

Authors: Irina Benedek, Oana Vanta

Keywords: Neurorobotics, robotic-assisted neurorehabilitation, robots, laws of robotics

This article comes as a continuation to  Neurorobotics: What should neurorehabilitation robots do for health practitioners? 

What are the fundamental laws of neurorobotics (robotic-assisted neurorehabilitation)?

The general regulations of robots provide a code of ethics for robots or human programmers of their artificial intelligence. Human health comes first in the hierarchical structure of these laws, then human will, and ultimately robot self-preservation [1]. 

Marco Iosa et al. [2] reformulated the three rules of robotics into three laws for robotics in neurorehabilitation following the previously indicated desirable characteristics of a neurorobot (Figure 1).

Figure 1. Laws of neurorobotics

• The first rule: “Need for High Benefit/Risk Ratio” 

A patient cannot be hurt by a neurorehabilitation robot or be exposed to risk.

Neurorobots should be safe from a medical standpoint and in terms of mobility. The definition of “harm” should be broadened to include any potential patient injury. Because the patient may use that time to receive a more successful therapy, time spent on a useless robot, only marginally helpful or even harmful, should be viewed as damage. So, according to the first law, robot usage should be at least as safe and effective as other therapies, which suggests that it should have a more excellent benefit-risk ratio than human treatments [2]. 

There is currently a lack of clear information on how to utilize neurorehabilitation robots regarding: 

• optimal usage
• treatment frequency and duration
• safety measures
• potential adverse effects, etc [2].

However, the success of a therapy (including one using a neurorehabilitation robot) is dependent on the patient’s characteristics [2]  described in Figure 2 below:

Figure 2. Patient characteristics

• Second law of neurorobotics: “Tool for Therapists” 

Robots must follow therapists’ instructions unless doing so would violate the First Law.

Thus, neurosurgery robots are more popular than neurorehabilitation robots because the former supports the surgeon rather than replaces him. Similar to this, a robot for rehabilitation should not be seen as a stand-alone device [3] but rather as a tool in the hands of therapists, providing them with more precise movements, more intense, repeatable, or adaptable patterns and also relieving them of fatigue. By talking to the patients, encouraging them, and collecting feedback on tiredness, discomfort, or mental stress (parameters challenging to detect with sensors), the therapist should be involved in the loop to push the symbiotic equilibrium between robot and patient towards an optimal level [2]. 

     Marco Iosa and the team proposed an optimal patient-therapist-robot loop (see Figure 3).

Figure 3. The patient-therapist-robot loop (Available from [2])

• The third law: “Artificial Intelligence as Support for Human Intelligence”

A robot must transparently adjust its behavior to the patient’s abilities as long as it doesn’t violate the First or Second Law. 

 Although the robot’s artificial intelligence should not be restricted to the safety control of human decisions, the presence of a therapist is necessary to aid the device. There are various factors to calibrate and adapt throughout recovery. Robot artificial intelligence should be able to automatically carry out all the control adjustments the therapist requires while giving them a clear quantitative picture of all these changes [2].

 A new generation of human-machine interfaces built into neurorobotics should be developed, allowing the robot to convert commands from the therapist at the macro level into micro modifications, following this set of rules [2]. 

What can be extrapolated from the field of neurorobotics?

• Most research and evaluations on robots for neurorehabilitation have concentrated on their efficacy but have found inconsistent results. 
• Robot ethics have received little attention because artificial intelligence is still a new concept. 
• Data suggest that patients and therapists are reticent to use robots in rehabilitation.
• The three suggested rules of neurorobotics emphasize the moral requirement to demonstrate a robot’s efficacy before commercialization and the ideal characteristics that neurorobotics should possess. 
• It is necessary to involve the therapist in the communication between the patient and the robot. 
• Neurorobots might be an essential instrument in the therapists’ hands, assisting them in repeated and rigorous patient movement and giving them quantitative data on a patient’s deficiencies, remaining skills, and functional recovery [1].

For more information on robot-assisted neurorehabilitation, visit:

• Robot-assisted neurocognitive rehabilitation of the hand
• The Effectiveness of Robotic-Assisted Therapy in stroke rehabilitation

References

1. Morasso P, Casadio M, Giannoni P, Masia et al. Desirable features of a “humanoid” robot-therapist. Conference proceedings: Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2009:2418–2421. DOI: 10.1109/IEMBS.2009.5334954
2. Iosa M, Morone G, Cherubini A, Paolucci S. The Three Laws of Neurorobotics: A Review on What Neurorehabilitation Robots Should Do for Patients and Clinicians. J Med Biol Eng. 2016;36:1-11. doi: 10.1007/s40846-016-0115-2.
3. Morone G, Masiero S, Werner C, Paolucci S. Advances in neuromotor stroke rehabilitation. Biomed Research International. 2014;2014:236043. Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4090496/