Robotic prostheses as products enhancing the rights of people with disabilities. Reconsidering the structure of liability rules

Abstract

The term ‘robot’ refers to a wide variety of devices, serving very different purposes. The case of robotic prostheses is considered here. After defining such devices, and briefly describing the technical peculiarities that characterize their functioning and distinguish them from traditional implants, their relevance with respect to the fundamental rights of people with disabilities is considered. Pursuant to the United Nations Convention on the Rights of Persons with Disabilities, a claim is made that favouring the development of said applications may be required to subscribing states. In light of such considerations, the liability regime – namely that emerging from the Defective Product Directive – is analysed, in the attempt to determine – with a prospective analysis – the impact that said rules are likely to have on their development. The technology-chilling effect and the anticipated extremely complex evidentiary burden the user would have to face in order to obtain compensation lead us to conclude that legal reform is advisable. Some alternatives are considered, and in particular that of the development of a – partially – state-funded no-fault plan, intended to allow more ex ante certainty for producers and researchers (favouring the development of said devices), and prompt and adequate compensation to the victim in case an accident occurs.

Keywords:

• persons with disabilities
• robotic prostheses
• product liability
• no-fault plans

Conflict of interest disclosure

No potential conflict of interest was reported by the author.

Notes

1. The well-known citation is found in the Declaration of Independence, Action of Second Continental Congress, July 4, 1776, The Unanimous Declaration of the 13 United States of America, also available at http://www.uscis.gov/sites/default/files/USCIS/Office%20of%20Citizenship/Citizenship%20Resource%20Center%20Site/Publications/PDFs/M-654.pdf (last accessed 15 April 2015).

2. For the sake of brevity, please see Bertolini (2013).

3. The person who invented the term ‘robota' is the Czech novelist Karel Čapek, in his 1922 drama (Čapek 1922), indicating work associated with submission or slavery.

4. The definition is taken by the Merriam Webster Dictionary, and is available at http://www.merriam-webster.com/dictionary/robot (last accessed 15 April 2015). Similarly, the one offered by the Oxford Online Dictionary, in which a robot is defined as ‘a machine capable of carrying out a complex series of actions automatically [ … ] (especially in science fiction) a machine resembling a human being and able to replicate certain human movements and functions automatically’, available at http://www.oxforddictionaries.com/definition/english/robot (last accessed 15 April 2015).

5. The very effective image of an Evil Deity offering mankind its gift, and demanding a price in terms of human lives, was developed by (Calabresi 1985) to address the issue of policy regulation of traffic accidents.

6. According to this perspective, and with respect to the first kind of movements, the bearing factors in the choice of the grip are rather represented by the amount of power required to have a firm hold and the degree of precision needed to perform the task for which the object is held in the first place (Napier 1956, 902–903). Evidence shows that the same shape may induce different grips depending on the action to be completed. Holding a pen or using it for writing induce very different grasp choices in individuals, as well as filing a flat part as opposed to a round one in a machine, see the considerations of Cutkosky (1989, 269). The shape of the object and its texture will also determine the amount of strength required in order to maintain the hold on the object (Taylor and Schwartz 1955). The posture of the forearm and of one hand with respect to the other may further complicate the kind of grips a person may use (Napier 1956, 906 ff.; Cutkosky 1989, 269).

7. For a more detailed discussion, see Ohnishi (2007, 44–45). In particular, younger adults who suffered a traumatic amputation require greater sophistication of prostheses than older people in the same conditions. This shows that the needs posed by society are evolving and so are the tasks the person desires to complete.

8. Commercial electric prosthetic hands are generally single degree of freedom devices, since they only allow the opening and closing of a clamp like tool. Otto Bock Healthcare, Italy, 2013 [http://www.ottobock.com] and for technical information see http://www.ottobock.com/cps/rde/xchg/ob_com_en/hs.xsl/6953.html (last accessed 15 April 2015).

9. The most common prostheses are body-powered, meaning that mechanical means such as cables are used to allow the transmission of a movement (biscapular abduction) to the artificial hand, causing it to close as a clamp around the desired object.

10. Non-invasive UPIs use various bio-signal interfaces, such as electroencephalogram (EEG, measured on the surface of the skull) or electromyography (EMG, signals captured from the muscle). In some cases, they may also resort to so-called sensory substitution feedback systems, as well as other kind of interfaces (namely joysticks, and various kinds of remote controls could be placed under the foot or tongue, or rather use speech activated commands as well as the eye movement) in order to direct the prosthesis. Low invasive UPIs include targeted muscle reinnervation (TMR), where a direct connection between the prosthesis and the remaining part of the amputated muscle is created, or the localized implant of myoelectric sensors. Invasive UPIs should instead imply the direct neural control of the cybernetic hand as if it were the corresponding natural body part (Ohnishi 2007, 45 ff.).

11. See Carmena (2012). Experiments conducted on monkeys appear to show that learning to control an artificial body part should not be different for the brain from learning to ski or swim.

12. The ability of a decoder to assess what the meaning of the biological signal received actually was varies as a function of the duration of the signal itself. If the incoming signal lasts 900 ms (milliseconds) the accuracy can be much greater than if it lasts only 300 ms. Yet such a duration is incompatible with a practical use in everyday life. The experimental studies conducted on monkeys show that an average accuracy of about 85% can be achieved, while in other cases there could be false positive signals. For the discussion of such data see Controzzi et al. (2012).

13. In the cases considered, in particular by referring to people with disabilities, attention is primarily devoted to the restoration of a lost – or eventually never possessed – function, rather than pure enhancement. The very distinction between the two can be theoretically challenged, however doing so falls beyond the purposes of the current study.

14. Technological development, while addressing some vulnerabilities correspondingly creates new ones that previously did not exist, exposing the user to new and at times unconceived risks (Coeckelbergh 2013). In the case of a robotic hand for instance, while the wearer may recover the ability to manipulate objects and drive, because of that very reason he may also be exposed to accidents he would not have been involved in otherwise. The improvement in the quality of life of the person with disabilities is substantial, hence desirable even socially, but it is not without costs, which instead are the natural consequence of the materialization of – some of – the potential new risks.

15. For some critical considerations see Bertolini (2013).

16. Autonomy is deemed one of the more relevant aspects in robot design, and some authors elevate it to the most distinctive trait (Santosuosso, Boscarato, and Caroleo 2012), see also Matthias (2010), and Asaro (2007). In fact, the ability of the robot to operate unattended represents one of the main purposes of developing such technologies in the first place, leveraging human capabilities and freeing up time for more challenging and less repetitive activities. Yet the notion of autonomy should be more narrowly defined since it may result in unacceptable misunderstandings (Gutman, Rathgeber, and Syed 2012).

17. Directive 85/374/EEC of the Council of 25 July 1985 on the approximation of the laws, regulations and administrative provisions of the Member States concerning liability for defective products, OJ 1985 L210/29.

18. Directive 90/385/EEC of the Council of 20 June 1990 on the approximation of the laws of the Member States relating to active implantable medical devices, OJ 1990 L189/17, that at Article 1 defines them [2nd paragraph, let. (c)] as ‘any active medical device [thus requiring a source of power other than that directly generated by the human body, see let. (b)] which is intended to be totally or partially introduced, surgically or medically into the human body [ … ] and which is intended to remain after the procedure’.

19. The adoption of a European Charter of Fundamental Rights in fact did not produce any federalizing effect, thus superseding the existing constitutional traditions of the single European Member States (Strozzi 2011).

20. For an overall discussion of the Charter's rights that could be interested by the development of robotic technologies see (Koops et al. 2013).

21. Beylevel and Brownsword (2001) point out at least two alternative conceptions of human dignity, either as an empowerment – justifying the recognition and protection of human rights and freedoms – or a constraint – to freedom of choice with respect to the completion of given acts and practices deemed to violate a collective good, namely the dignity of others and of Man itself (see also Rubin 2009, Bostrom 2009).

22. For a more detailed discussion of this aspect see Koops et al. (2013).

23. The European Union itself ratified the convention on 21 January 2011, for an analysis on the status of the implementation of the convention, see the Commission staff Working Document on ‘Report on the implementation of the UN Convention on the Rights of Persons with Disabilities (UNCRPD) by the European Union’, available at http://ec.europa.eu/justice/discrimination/files/swd_2014_182_en.pdf (last accessed 15 April 2015).

24. Which may be seen as a mere obligation of conduct, see Quinn (2009b, 99 ff.), but may instead justify a more active normative intervention by the European Union, as described below.

25. For a general discussion on the rights and duties as arising from the UNCRPD, see Quinn (2009a), Bongiovanni (2011).

26. See the ‘Report on the implementation of the UN Convention’, cit., supra nt.24, 10, where at Section 31 it is stated that the right set forth by art. 4 let. g., was enacted ‘Through various programmes, most recently the 7th Framework Programme for Research and Technological Development for 2007-1319, the Commission has funded research projects on new technologies, including information and communications technologies (ICT), mobility aids, devices and assistive technologies and transport, and on socio-economic issues addressing the daily needs of persons with disabilities and the elderly’.

27. See the seminal article of Coase (1990), as well as Calabresi and Melamed (1972) and Calabresi (1985).

28. A measure that is anyway extremely necessary to enable the active participation of people with disabilities in society, for a discussion see Quinn (2009c, 41 ff.).

29. The estimated impact of robotics for healthcare – among which prostheses and exoskeletons – on the World economy by 2025 amounts to $800 billion to $2.6 trillion a year, see Manyika et al. (2013, 68). Much of this overall effect would benefit users directly by increasing the quality of the life (an estimated effect of $390.000 per person was assessed making use of a QALY approach, Manyika et al. 2013, 74). This is also due to the more active participation of people with disabilities in society, dramatically changing existing conditions. For a discussion see Bertolini and Palmerini (2014, 98 ff.).

30. For a discussion on this subject matter see also Schellekens (2014).

31. An exception is granted for custom-made devices, which still need to meet the same standards of safety but may be commercialized so long as the manufacturer draws a declaration conforming to annex 6.

32. Specified as the ability of the device to bear the stresses which may occur during its use, providing an advantage that makes collateral risks acceptable (see annex 1, §I general requirements).

33. In this direction could be interpreted the defence affirmed by art. 7, §1, let. d) DPD. The existence of a normative standard is always meant to be a minimum level of safety required, thus it does not exclude the possibility to affirm liability for damages resulting nonetheless from the use of the thing. See for instance Cass., 20.7.1993, n. 8069, Foro italiano, 1994, I, 459. In the State of NY, compliance to industry standards and governmental regulation may be used to prove the absence of defect, see Voss v. Black &Decker Mfg. Co., 59 N.Y.2d 102, 450 N.E.2d 204 (1983), but pursuant to George v. Celotex Corp., 914, F.2d 26, 28 (2d Cir. 1990) industry practice may not be sufficient, for it may be lagging behind in technological development.

34. Art. 13 DPD saves the peculiarities of each legal system by stating that ‘This directive shall not affect any rights which an injured person may have according to the rules of the law of contractual or non-contractual liability or a special liability system existing at the moment when this directive is notified’, and this aspect will inevitably add onto at the variety of the overall context with which the rules set forth by the directive will have to be coordinated – on this aspect see Castronovo (2006, 659).

35. Moreover, provided that pursuant to art. 7, §1, let. b the producer may be liberated by proving only that it is probable that the defect did not exist at the moment the good was commercialized, it is correct to assume that the opposite is also true. Therefore, for the claimant to ground his case the demonstration that it is probable that the damage occurred as a consequence of the use of the good may as well suffice, see Castronovo (2006, 699). However, see the more recent decision by the Italian Corte di Cassazione, Cass., 8.10.2007, n. 20985, in Danno e Responsabilità, 2008, 290 ff., and the discussion by Pardolesi and Bitetto (2008), Querci (2008), requiring the user to prove defectiveness specifically. In NY defectiveness may instead be proved indirectly by showing that the product did not perform as intended and that all other reasonable causes of the accident – not depending on product defectiveness – may be excluded, see for instance Ramos v. Howard Indus. Inc, 10 N.Y.3d 218, 223–24, 885 N.E.2d 176, 178–79 (2008).

36. See for instance Cass., 25.5.1964, n. 1270; Cass., 22.11.1995, n. 12023 where the inversion is obtained by the so-called presunzione semplice set forth by art. 2729 c.c.; and for German courts see BGH, 30.4.1991, in NJW, 1991, 1951. For some reference to Spanish case law see Torres Mingot (2005, 293 ff).

37. The defences of Article 7, lett. a, b, c, make this point excluding liability for every case where the assumption of risk is lacking.

38. See also Indovino (2008). This is also true for many US jurisdictions. Renouncing the ambition of completeness, see for instance George, cit. supra nt. 34, at 28, stating that a manufacturer is held to the knowledge of an expert in its field as a duty ‘to keep abreast of scientific knowledge discoveries, and advances and is presumed to know what is imparted thereby’. In other jurisdictions such as California, it is taken into account when discussing the feasibility and cost of alternative designs, see McLaughlin v. Sikorsky Aircraft, 148 Cal. App. 3d 203, 195 Cal. Rptr. 764 (1983).

39. The possibility to exclude this defence is contemplated by art. 15, §1, let. b) DPD.

40. See Castronovo (2006, 708). Under French law, the transposition of the directive stated that the development risk defence could not be invoked in case the problem with the product arose or was discovered within 10 years from its introduction into the market, and the producer failed to take adequate measures in order to prevent harm. This limitation of the defence was deemed illegitimate by the European Court of Justice, which so ruled on April 2002. Yet the same limitation may be applied by resorting to normal French liability rules, as stated by Le Guillou and Bresson (2005, 250–251). For Germany see Foerste and Graf von Westphalen (2012, 834–835).

41. In some European countries a tendency can be observed to sue under tort law together with product liability rules (Best 2002, 4).

42. The norm however, sets a rather high cap for compensation, by stating it cannot lower 70 million ECU. Germany has provided for liability caps for both normal products and drugs, abiding different regulation.

43. Also known as Risk Utility Test, see for a brief discussion Geistfeld (2011, 115 ff.); Foerste and Graf von Westphalen (2012).

44. Italian case law seems to intend the notion of foreseeable use in a quite strict fashion, allowing recovery only in case a rationale and therefore predictable use of the good is made. For instance in a case decided first by the Court of Appeals of Genoa, 28.3.1992, confirmed by Cass., 29.11.1995, n. 10274, compensation was denied since the abnormal use of a swing made by a 12-year-old, who stood on top of the seat rather than sitting down on it, was deemed not rationale, thus not foreseeable. Other national courts could in this respect be less lenient towards the producer. The degree of rationality of a child shall in fact be anticipated by the manufacturer of swings and he could expect the product to be used in an inappropriate fashion, and thus design necessary safeties. See also Foerste and Graf von Westphalen (2012, 863) and Best (2002) commenting on the British case A & Others v. The National Blood Authority.

45. Owen and Phillips (2005, 171 ff.), in particular under US law it is important to correctly interpret comment j to §402A of the Restatement Third of Torts: Products Liability (henceforth Rest.), which does not discharge – per se – the producer who duly informed about the inherent risks of the product. Under Italian law see Busoni (2007, 838).

46. See supra nt. 45, (Busoni 2007, 841–42). In California, both the consumer expectations test and the risk-utility test are applied, see Baker v. Lull Eng'g Co., 20 Cal. 3d 413, 143 Cal. Rptr. 225 (1978).

47. Art. 1 DPD, which holds the producer liable for the damage caused by a defect in his product, is intended as not limited to the person owning the good, but rather includes even those who make an unauthorized use of it, see Busoni (2007, 834).

48. This was the case in the United States with the general aviation industry, where an exemption from the application of normal product liability rules was introduced for aircraft 18 years or older. In 1994, the General Aviation Revitalization Act (Act Aug. 17, 1994, P.L. 103–298, § 1-4, 108 Stat. 1552; Nov. 20, 1997, P.L. 105–102, § 3(e), 111 Stat. 2215, henceforth GARA) was enacted. The study by Helland and Tabarrok (2012) shows that the number of accidents did not increase, and actually declined. For a discussion of the implications and differences with robotic prostheses, see Bertolini (2014a).

49. Like the one introduced by the GARA, see supra nt. 49.

50. Current standards have instead a very broad range of application. The AIMDD applies to anything ranging from a pacemaker to a robotic prosthesis, and other relevant directives, such as the Directive 2006/42/EC of the European Parliament and of the Council of 17 May 2006 on machinery, OJ 2006 L 157/24 (amending Directive 95/16/EC), (henceforth MD), to any ‘assembly of linked parts or components, at least one of which moves [ … ]’ (art. 2 MD). For a discussion see Bertolini and Palmerini (2014, 120).

51. Intentional torts irrespective of whether the limb was involved or even malfunctioned should not fall under the plan, shielding the wearer from liability actions brought against him by injured third parties. Compensation through the fund to such third parties may still be ensured if the use of the limb was involved, malfunctioning could not be radically and completely excluded and only in a subsidiary position should the tortfeasor be insolvent.

52. Meaning that there is radically no causal nexus between the use of the limb and the harm caused, for instance because only the natural limb was used in order to cause harm (the implantee hit even non-voluntarily a third party with his natural limb) or the harmful conduct was anyway totally unrelated to the use of the prosthesis (as in cases of negligence of the wearer).

53. With respect to non‐pecuniary losses, it may be conceived that only relevant prejudices to bodily integrity are liquidated, but not other forms of pain and suffering or other damages, that could otherwise be demanded in a trial, pursuant to existing regulation and case law in each single MS.

54. The producer may be held liable, and thus sued in court without the possibility to benefit from the exemption, should gross negligence (or scienter) be ascertained through a prima facie assessment, once the compensation claim is filed with the competent authority.

55. Indirectly they may also contribute to the extent that the prosthesis implanted was purchased by the national healthcare system of the member state.

56. As happens with traffic accidents regulation (Calabresi 1985).