From autonomous cars, robotised hotel complexes to surgical and companion robots, the technology of robotics have well advanced. However, have the regulatory measures followed with the same pace? We believe that it has not evolved fast enough. At present, only South Korea has adopted a law that specifically governs the sector of robotics whilst other jurisdictions still struggle how to address immense technological advancements and the related challenges from a legal perspective.
At this point, robots have not yet taken the main stage at the global market but, nevertheless, studies show that this will steadily change1, given the benefits that robots could bring to our society. Robots may be used to decrease road fatalities, ensure a longer autonomous life of elderly, improve health, reduce repetitive tasks of care personnel, ensure easier operations in disaster areas, increase higher productivity of industry, etc. Thus the advancement of robots is unstoppable and it is necessary for crucial aspects like safety, liability, and data protection to be regulated. The debate around employment policies, the so-called robot tax2, and the universal income3 could be added. However, the latter remains politically disputed and surpasses the limit of this article.
Accordingly, this article presents the international framework, particularly focusing on the recently adopted report by the European Parliament (I.), an overview of soft law standards (II.) and the national reference legislation and strategies on robotics (III.). This part is followed by examples of regulatory techniques that could be included in future robotics’ legislation (IV.) and some links (V.) that might be useful to the reader.
I. International guidelines and supra-national law
1. European Union Law
The law of the European Union (EU) does not contain specific rules on robotics, however there is EU legislation that might be applicable to certain robotic products:
furthermore, there is the Machinery Directive which applies is applicable to many types of robots;
the Product Liability Directive that could equally apply to automated vehicles.
However, those directives however do not cover many of the specific and complex issues related to intelligent robots, such as the autonomous decision-making and human-robot interactions.
The European Parliament (EP) recently opened a debate in this respect calling for an EU-wide legislative action. The EP Report, adopted in February 2017, focuses on civil law rules on robots and artificial intelligence. The report calls on the European Commission (EC) to present relevant legislative proposals to fully exploit the economic potential of robotics and to guarantee a standard level of safety and security. The EC is called upon to address issues of liability, safety, privacy, as well as the impact of robotics on employment, health and environment.
The EP report differentiates among different types of robots, e.g. care robots, medical robots and autonomous means of transport, i.e. autonomous vehicles and drones. It envisages that the care robots shall focus on automated tasks, while permitting medical staff and caregivers “to devote more time to diagnosis and better planned treatment options”. When it comes to medical robots, the report requests trainings of health-care professionals and the creation of robotic ethics committees within the hospital. For drones and automated vehicles, aspects such as safety, security, privacy, civil responsibility and tracking systems shall be addressed.
Generally, the report encourages to further strengthen research in robotics by scaling up the funding.
To avoid disparities within the Union’s internal market the A classification list, testing, certification, market approval, registration and market surveillance regarding the advanced robots shall be established at EU level to avoid disparities within the Union’s internal market. The EP equally advocates for the creation of European Agency for robotics and artificial intelligence. The report calls for international harmonisation of standards and of testing of robots in real-life situations. Moreover, production of robots shall be environmentally friendly.
When it comes to privacy and data protection, a special attention shall be placed on robots operating in private spheres. To reflect specific aspects of robotics, an update of existing copyright legislation is suggested while respecting GDPR (General Data Protection Regulation) and calling for a high level of security of data flows, i.e. security by design.
Finally, several liability schemes are proposed, namely the compulsory insurance scheme and a compensation fund. An instrument for collective claims of consumers is equally suggested.
2. Regulatory guidance: RoboLaw
The EU funded RoboLaw project, closed in 2014, provides guidelines on ways of regulating robotics, while addressing four different application areas: driverless vehicles, robotic prostheses (and exoskeletons), surgical robots and robot companions. It points out “the technical peculiarities of each, and identifies both the ethical and legal implications that the emergence and diffusion of a similar technology may give raise to”.
Contrary to regulators, standardisation organisations have already heavily invested into robotics.
a) International Organization for Standardization (ISO)
Standards adopted by the ISO are both the most universal and the most sector specific. These standards can be purchased from the national standardisation organisations which are members of ISO. The standards define the main terms used in the sector of robotics as well as the requirements for designing, safe operation of industrial, personal care and collaborative robots. The main standards that have already been put in place are:
ISO 8373:2012 Robots and robotic devices: vocabulary
“It defines terms used in relation with robots and robotic devices operating in both industrial and non-industrial environments.”
ISO 10218-1 Safety requirements for industrial robots — Part 1: Robots
“It specifies requirements and guidelines for the inherent safe design, protective measures and information for use of industrial robots. It describes basic hazards associated with robots and provides requirements to eliminate, or adequately reduce, the risks associated with these hazards.”
ISO 10218-2:2011 Safety requirements for industrial robots — Part 2: Robot systems and integration
“It specifies safety requirements for the integration of industrial robots and industrial robot systems as defined in ISO 10218-1, and industrial robot cell(s). It describes the basic hazards and hazardous situations identified with these systems, and provides requirements to eliminate or adequately reduce the risks associated with these hazards.”
ISO 13482:2014 Safety requirements for personal care robots
“It specifies requirements and guidelines for the inherently safe design, protective measures, and information for use of personal care robots, in particular the following three types of personal care robots:
mobile servant robot;
physical assistive robot;
person carrier robot. ”
ISO/TS 15066:2016 Safety requirements for collaborative robots
“It specifies safety requirements for collaborative industrial robot systems and the work environment, and supplements the requirements and guidance on collaborative industrial robot operation given in ISO 10218-1 and -2.”
b) Institute of Electrical and Electronics Engineers (IEEE)
IEEE is a technical professional society – an association dedicated to advancing innovation and technological excellence. It is designed to serve professionals involved in all aspects of the electrical, electronic, and computing fields and related areas of science and technology that underlie modern civilization4.
IEEE Standard Ontologies for Robotics and Automation
“Standards defining activities in both established, mature application areas and nascent, emerging technologies related to robotics and automation.”
III. National reference legislation and strategies
1. General trends and issues
It seems that only South Korea so far has put in place a specific law on robotics. Japan, UK and the United States have recently adopted their national strategies / roadmaps in the field which, in the case of Japan, includes proposals for whereas Japan also suggesting some regulatory reforms.
Generally In most cases jurisdictions identify certain different categories of robots, such as automated vehicles, drones, assistive robots, medical/surgical robots and related highly intelligent machines. These jurisdictions intend to regulating in a specific way for separately these categories. We believe that this is for the time being the appropriate approach, as different robot categories imply different risks and accordingly demand divergent levels of safety and security. However, it might be possible to define cross-cutting regulatory requirements over time. If so and if ever new categories of robots appear, it might be more appropriate to establish generic regulation in addition to the category-specific regulation.
In line with the prevailing category-specific approach, many argue that liability schemes should be category-specific and should put unlike differentiated degrees of responsibility on designer / manufacturer / producer / user, especially in view of the unpredictability of robots’ actions and the non-foreseeable damages.
The main questions that arise are:
Who shall be the responsible party?
How safe shall the robots be and by which measures shall safety be ensured?
By which measures can robots be protected against hacking?
How to ensure privacy?
2. South Korea
South Korea adopted a law on robotics in 2008, which is so far – to our knowledge – the only country with such a legal framework. South Korean law aims at promoting the development and distribution of intelligent robots whilst laying down general rules. These rules, which shall be further implemented by a 5 year governmental “basic plan” every 5 years. According to this plan, the government shall adopt more detailed rules regarding the mid-term and long-term goals, promote the robot science and the practice of the “Robot Ethics Charter” (see below). The latter represents a “code of conduct established for persons involved in the development, manufacture, and use of intelligent robots in order to prevent various kinds of harmful or adverse effects, such as destruction of social order, that may arise from the development of functions and intelligence of intelligent robots, and to ensure intelligent robots contribute to enhancing the quality of life of human beings”.
The South Korean law defines the term of “intelligent robot” as a “mechanical device that perceives the external environment for itself, discerns circumstances, and moves voluntarily”. Incentives such as financial support, loans or any other administrative support are foreseen for a person who undertakes any project to develop and distribute intelligent robots. Moreover, the South Korean law has instituted and defined special “Intelligent Robot Investment Companies” that shall invest in the development and distribution of finished products or parts of intelligent robots and technology. In this respect, these companies benefits from tax reductions and exemptions in order to boost private investments. Equally, a joint penal responsibility is foreseen, of at both the individual and the company levels. Moreover, the law creates and thoroughly regulates “Robotland”, i.e. an area(s) where facilities for the utilisation of various kinds of intelligent robots and other auxiliary facilities are installed for the development and distribution of intelligent robots. These areas might be developed as a Public Private Partnership,, might be subsidized by the government, etc. including the possibility to collect entrance fees.
Commendably, the law foresees wide accessibility of robots, i.e. free usage of robots by socially disadvantaged people, such as the elderly, and low-income earners. When it comes to safety, security and data protection, the law refers to the “Robot Ethics Charter”, which is still to be enacted. The latter represents a “code of conduct established for persons involved in the development, manufacture, and use of intelligent robots in order to prevent various kinds of harmful or adverse effects, such as destruction of social order, that may arise from the development of functions and intelligence of intelligent robots, and to ensure intelligent robots contribute to enhancing the quality of life of human beings”. Some suggest that the Charter will take as a basis the famous science-fiction Asimov’s “Three Laws of Robotics”:
A robot may not injure a human being or, through inaction, allow a human being to come to harm.
A robot must obey orders given it by human beings except where such orders would conflict with the First Law.
A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
Additionally, the minister has now an obligation to establish policies to secure the quality of intelligent robots as the whole chapter on quality certification system was deleted in January 2016. Nevertheless general legislation on industrial standards remains to apply for already certified robots.
Furthermore, a special Institute for the promotion of robotics’ industry advancement is established, enumerating a long list of its functions. Finally, promotion of international cooperation is envisaged by the following means:
Surveys and research for international cooperation;
International exchange of human resources and information;
Holding exhibitions and scientific conferences;
Activities for the international standardization under the Industrial Standardization Act;
Overseas marketing activities, attraction of foreign investments, and public relations activities;
Invitation of foreign institutions or organizations related to international conferences.
South Korea has adopted an interesting law on robotics and is a good beginning for regulating this issue. We believe that it will develop over time and shall include more stringent provisions, namely on safety and security aspects.
Japan’s Robot Strategy aims at turning the country into “a showcase for robot utilisation” by supporting technological development, encouraging introduction of robots all over the country, and improving market environment. In this respect, deregulation and a creation of new legal framework is deemed to be needed. We present certain changes suggested by the Strategy:
Robotic medical/surgical devices: review period from the receipt of application of a brand-new medical device to its approval are to become 14 and 10 months for normal review item and priority review item, respectively.
Care robot devices: necessary to accelerate their diffusion by providing appropriate support for developing companies and care workers. The coverage of the long-term care insurance system should be more flexible.
Electric personal assistive mobility devices, unmanned farm machines: to update regulation for them to be admitted to public roads while respecting safety verification.
The Strategy equally envisages to strengthen measures for securing consumer safety considering the overall presence of next-generation robots’ in every aspect of our life, especially taking into account robots’ autonomy, remote controllability and information collection. When serious accident happens due to a robot, a range of responsibilities of manufacturers shall be considered.
Moreover, a Robot Award shall be expended to further accelerate efforts by individuals toward the realisation of robot revolution. Additionally, Robot Olympic Games shall be held in 2020 to compete in robotic technology and to solve “actual issues in various areas, such as medical and health care, infrastructure inspection, agriculture, forestry and fisheries industry, manufacturing industry, service industry, and entertainment industry; it makes actually useful robots compete each other; and it shows to many people what robots are.“ The Strategy also calls for raising public awareness, support to SMEs and the reinforcement of international cooperation in setting up rules on robots, including those regarding their safety.
4. United States
US Roadmap for robotics is a technical roadmap laying down milestones to be reached within the next 5, 10, and 15 years. The Roadmap only briefly addresses certain legal challenges, i.e. safety, liability, privacy and security of robotics and puts the human-robot interaction (HRI) research as a top priority.
5. United Kingdom
In February 2017 the UK published its national Digital Strategy, while recognising that a strong, stable and innovation-friendly regulatory regime is a must for the sector of new technologies.
6. Specific categories of robots
For specific categories of robots, we suggest studying the following regulation:
Automated vehicles are the most widely regulated category within the sector of robots and artificial intelligence. 11 of the United States have already passed the relevant legislation, as well as certain EU countries (UK, Germany – draft legislation). The European Commission is currently reviewing the existing regulatory framework and will decide on whether to issue an EU legislative proposal in 2018.
For industrial robots Japan adopted an ordinance that lays down rules on training of workers, inspections, checkups and prevention of dangers. Equally, US has set up guidelines for robotics safety in the manufacturing sector.
When it comes to robotic prostheses the authors of the Robolaw project argue that they shall be recognised as products and when incorporated, forming a part of a human body. High technical standards regarding safety, e.g. for complex control systems/ brain – computer interface sending brain generated signals to muscles and adoption of certification systems are necessary. Liability should not be too strict because otherwise it would hamper research and improvements. The authors recommend the establishment of a compensation fund to cover damages for wearers and third parties. The fund shall not require any proof of fault, but compensation shall be capped at a certain limit.
Assistive robots might need a legal guardian, and this in particular in view of the conflict of interests, i.e. among the relatives and in relation to the insurance company5. Additionally, questions of sensitive data protection and protection of personal liberty choices are very relevant in this category of robots. Imagine that the personalised alarm reminding to take a medicine was hacked or personal data were disclosed without explicit authorisation instead.
IV. Missed regulatory techniques
As legislation on robotics is so much in its infancy, we recommend incorporating regulatory techniques from other regulatory sectors. Many of them have been identified in the Handbook – “How to regulate?” (see right column of this website).
1. Risks and performance requirements
Dealing with risks and performances is key in the sector of robotics. Regulators can try to fix a limit of acceptable risk, either qualitatively or – at least in theory – quantitatively, i.e. to require that a certain hazard does not become reality in more than one out of 1000 or one out of 1000000 cases. Evidently, the acceptability limits should depend on the severity of the harm / hazard.
We believe that potential damage of robots is high, thus it might be a wise legislative decision to require full safety principle, i.e. decreasing risks to close to zero. In case of robots that bring along benefits, it might be more appropriate just to oblige to reduce risks as much as possible and to request a positive risk-benefit balance when i.e. the benefit is weighed against the sum of the risks (all types of risks combined). For robots claiming a benefit, it might be furthermore appropriate to require a certain level of performance, ensuring that the benefit can indeed be reached.
In order to adapt risks to technical progress, regulators can try to build in a kind of automatic adaptation by using a variety of techniques, e.g. expressions that imply an automatic update like “state of the art”, “best available techniques”, “technically most advantageous”, “to reduce as much as possible the risk …”; “… comply with to the most recent international standards”, etc.
More on Risks and performances can be found in Chapter 3. of Handbook – “How to regulate?”.
2. Robots sold as kits
Product legislation does often not cover kits. Newly conceived legislation on robots should not make this mistake. Kits intended to be built / configured to become robots should be covered from the beginning.
3. Robots not sold, but directly put into service
Classic product legislation applies to products sold / placed on the market and leaves a big loophole regarding products rented or otherwise put into service without purchase contract. To avoid the loophole, future legislation might refer to another reference point for establishing legal requirements different from the sale or placement on the market. Legislation might instead refer to the physical creation or to the putting into service / first use of a device (as some legislation already does today).
4. Robots in use
As robots require maintenance and sometimes even training of the user to be safe, legislation shall equally set up requirements for robotic devices in use. Besides the already mentioned aspects of maintenance and training of the user, requirements might sometimes be necessary to verify periodically the fitness of the device, to update its software, to protect it continuously against hacking, viruses and other malware. These obligations should be applicable to the economic operators providing access to the robotic devices, but be incumbent to the users as well.
5. Covering distribution by e-commerce from third countries
The distribution of robots will sometimes happen by means of e-commerce. E-commerce raises specific enforcement problems. Economic operators can be located anywhere in the world. Enforcement of product legislation is thus extremely difficult. The state/country, where the business is conducted, is not necessarily willing to help. And even if the authorities are willing to help, they will not necessarily have the legal empowerment to enforce product legislation requirements of other jurisdictions. Hence it is of paramount importance to include, in the legislation on robotics, the empowerment to block websites of operators in third countries if these operators do not fulfill the legal obligations. Furthermore, it would be useful to oblige all manufacturers, importers and distributors to display information on the undergone conformity assessment of the jurisdictions in which they intend to sell their regulated products.
6. Robust enforcement powers against trading operators
As it is often difficult to enforce product legislation directly in contact with the manufacturer, it is necessary to provide for empowerment to request from importers, wholesalers, retailers and trading platforms that they only offer robots from manufacturers who make accessible authorisations, certificates and other proof of conformity of their products (e.g. proof of registration in case of a registration obligation, proof of labelling in case of a labelling obligation, instructions for use).
7. Alert portal
Enforcement authorities are ever more struggling with capacity constraints. An alert portal should be available for whistle-blowers and clients who consider they detected non-conformities.
8. Provisional orders
Regulation can provide for an explicit empowerment to make provisional orders to avoid imminent danger. Section 4E of the Air Navigation Act of Singapore could serve as an example.
9. Liability and insurance
Whilst some authors have already mentioned the need for liability provisions applicable to designers / manufacturers / owners / users of robots, little has been said about the necessity to require insurance covering liability claims. However, it goes without saying that liability provisions are worth nothing in case of insolvency.
10. Classic requirements for products and services
Whilst robots are generally seen as something completely new, they are in essence machines. Machines exist for centuries, and legislation regarding machines – since the 19th century. Hence it is only natural that there are many elements of the checklist for products and services contained in the Handbook – “How to regulate?” (see right column of this website) that could be used for robots (and their putting into service) as well. We present them in the Annex.
Whilst we claim that these checklists will be very helpful to develop appropriate requirements for robots, we do not regard them as sufficient. These checklists do not cover the specific aspect of autonomous decision-making of robots and their interaction with humans. In that sense, it is advisable to complement the checklists, e.g. with specific aspects set out in the above mentioned international standards. To list but a few of these:
– Hazards due to incorrect autonomous decisions and actions ;
– Hazards due to lack of awareness of robots by humans ;
– Hazards due to localization and navigation errors ;
– Safety-related environmental sensing.
If you want to investigate further:
– Robolaw project ;
– Law and Technology Knowledge Engine is an open source of papers and articles worldwide on Information Technology Law.
This article has been written by Ajda Mihelčič, M.A.S., on behalf of the Regulatory Institute, Brussels.
ANNEX – checklists for products and services contained in the Handbook “How to regulate?”
“11.2.4 Particular checklist for regulation on products, objects (also partly applicable to materials)
Check whether you also need to regulate with regard to the following elements:
Product components, replacement parts;
Services used in the production process;
Services with regard to products once they are placed on the market;
Services with the products (offered to final users, consumers etc.);
Long-distance sales (e.g. via Internet), with special focus on salespersons outside the jurisdiction;
Manufacturing as an activity;
Professional activities linked to the products and their distribution;
Fees for conformity assessment activities;
Fees for enforcement / market surveillance activities;
Fees for the application of conformity assessment bodies as Conformity Assessment Bodies.
Check whether all of the following risks, if relevant, have been covered:
– Mechanical risks (e.g. failing of brakes, failing of steering, squeezing mechanisms, cutting mechanisms),
– Software failure risks,
– Risks of software manipulation,
– Risks of electric failure,
– Risks linked to electricity,
– Risks of incompatibility of devices, connectors, etc.
– Risks linked to electro-magnetic radiation (risk of interference with devices, risks for ultra-sensitive persons),
– Risk of radioactivity,
– Risk of other tissue destructing radiation (e.g. by protons or other parts of atoms),
– Risk of optical disturbance by beams and other light(s),
– Risk of too high or too low temperature,
– Risk of fire,
– Risk of too high pressure (e.g. in case of explosion),
– Risk of not performing sufficiently / as intended (e.g. medicine),
– Risk of misunderstanding instructions for use,
– Risk of unintended inappropriate use,
– Risk of intended inappropriate use (“off-label use”).
Check whether the right risk management principles have been applied to each of these risks.
Check whether you need specifically regulate on steps before the final product leaving the manufacturer, e.g. with regard to:
– Risks of research.
– Development of intermediate products.
– Making available of intermediate products.
– Development of final products (thus not the risk of the final product itself, but the risks of developing final products).
– The only sector known by the author for which there is, in some jurisdictions at least, a complete control, from research to final products, is the sector of nuclear research, distribution of nuclear materials and final nuclear products.
11.2.5 Particular checklist for regulation on services (also partly applicable to processes in general)
Check whether you also need to regulate with regard to the following elements:
Services used to provide the services;
Products used to provide the services;
Remedying deficient services;
Long-distance services (e.g. via Internet), with special focus on service providers outside the jurisdiction;
Professional activities linked to the services or linked to the distribution;
Service providing as a professional activity;
Fees for conformity assessment activities;
Fees for enforcement activities;
Fees for the application of conformity assessment bodies as Conformity Assessment Bodies
1 McKinsey study showed that by 2025 RAS technologies will have an impact on global markets of between $1.9 and $6.4 trillion per year.
2 “an obligation to pay for retraining of workers displaced by robots”
3 “a basic income for every worker subsidised by the gains in productivity and efficiency from automation”