Care robots: a gap in the market?

Assistive technology suppliers are very dependent on procurement by care provider organisations rather than a consumer market, which remains very small. As a result, suppliers emphasise cost savings for public services or care insurance budgets, and this is where much of the research effort has been concentrated. For example, the York Health Economics Consortium has estimated that a Scottish Government investment of £6.8 million in developing telecare delivered £11.2 million of savings in one year due to quicker hospital discharge, less unplanned admissions, reduced sleepover care and less home check visits (Beale, Sanderson and Kruger Reference Beale, Sanderson and Kruger2009). Evidence of benefit extends across falls prevention, less pressure on carers, better compliance with medication, and avoidance of dangerously low indoor temperatures (Yeandle Reference Yeandle2009).

In the UK there is now a focus on large-scale adoption following major demonstration initiatives such as England's Whole System Demonstrator (WSD) programme, running since 2008 and involving more than 6,000 participants, and Scotland's telehealthcare demonstration programme, starting in April 2012 and involving at least 10,000 older and disabled people (Sweet and Down Reference Sweet and Down2011). Other research initiatives concerned with mass uptake include a recent UK Technology Strategy Board call for research proposals, this time focusing on developing economic and business models to grow the telecare and telehealth markets, and understanding better the social and behavioural issues involved. The Board described a paradox of rapid technical development contrasting with slow market pull (Technology Strategy Board 2010).

While market pull has been slow, product push is very much in evidence but to public-sector procurers rather than individual consumers. The availability of product information for organisational purchasers is growing, with the UK's ‘Buying Solutions’ web portal listing over 900 telecare, telehealth and telecoaching products (www.buyingsolutions.gov.uk) and a dedicated website now supporting telecare and telehealth product comparison (www.alvolution.co.uk/compare/product_comparison_website). These sites do not list any care robots, and although care robots exist they are far outside the mainstream of telecare and telehealth products because they are costly and still largely experimental. This means that a product push for care robots based on saving money is not evident, contrasting with other spheres. Industrial robots, for example, now have a 50-year history, and there has been substantial growth over the last decade in robotically assisted surgery and physical and occupational therapy (Okamura, Matarić and Christensen Reference Okamura, Matarić and Christensen2010). In addition to these medical robots, the robotics market includes robots for manufacturing, service robots such as for cleaning or security, and entertainment robots. Over the past 20 years prices have fallen sharply while quality has continued to improve, along with new capabilities such as human–robot co-operation (Kumar, Bekey and Zheng Reference Kumar, Bekey, Zheng, Bekey, Ambrose, Kumar, Sanderson, Wilcox and Zheng2006).

The research and development effort in care robots is pursuing the same course as for telecare and telehealth, based on the promise of reducing labour costs and delaying admission to residential care. Funding from the European Commission is currently supporting major research and develoment projects such as LIREC (LIving with Robots and intErative Companions: http://www.lirec.eu) and SERA (Social Engagement with Robots and Agents: http://project-sera.eu). As with telecare equipment, these robots are often promoted as ways of managing age-related ‘decline’ rather than as labour-saving devices or lifestyle enhancements, terms more typically associated with domestic appliances and personal technology such as vacuum cleaners and smart phones. Yet it is these comparisons with ‘ordinary’ – albeit innovatory – consumer products that suggest a different course for assistive technology that could be pioneered by care robots.

Not only are telecare and telehealth products not marketed on any significant scale directly to the public because suppliers see their prime markets to be care organisations aiming to save money, but public attitudes also tend to be that supplying such equipment is a matter for public services and insurers rather than their own disposable income and ordinary shopping (Ipsos MORI Social Research Institute 2011; PricewaterhouseCoopers 2009; Yeandle Reference Yeandle2009). This is increasingly likely to be a problem when governments around the world are attempting to encourage their citizens to invest in their own care so that public funding can be concentrated where it is most needed as their populations age. For the majority of older people and their families there is no reason why buying something that helps with care should not be a normal consumer purchase. While it has not yet been possible for telecare equipment to break into a mass consumer market, similar functionality but packaged in a small robot may have the potential to make it to supermarket shelves. This would not be as a special need aid – as telecare, designed for and predominantly purchased by public services and professional care organisations, is usually thought of – but as a desirable and useful consumer product. Affordability is a challenge, but if development is refocused on certain specific functions that do not involve the costs involved with capabilities such as fetching and carrying or mimicking a human being or pet, and if the robot is embedded in social media, then both affordability and marketability are more likely to be achievable.

Commercialisation risk has been a barrier to innovation in assistive technologies generally. In 2007 the Foundation for Assistive Technology studied 362 assistive technology projects in small and medium-sized enterprises (where most innovation sits) and found that only 12 had delivered a commercial product (ATcare 2009). Issues included lack of research time, funding gaps, lack of commercial expertise and the market being underdeveloped. A key problem was the challenge of marketing a ‘really new product’. Really new products present consumers with products that are like nothing they have seen or experienced before (Lehmann Reference Lehmann1997). So, if care robots were to appear on the market, what would they actually be able to do?

Robot capabilities

Robotic platforms are now generally available at relatively low cost. A small robot with on-board processing, radio communication and a mobile base platform can be built from commercial off-the-shelf components for around £100. The Roomba robot vacuum cleaner is essentially a low-cost platform, retailing at £200–300, and is able to support much more than just vacuuming tasks (Tribelhorn and Dodds Reference Tribelhorn and Dodds2007). Localisation, mapping and navigation capabilities are improving all the time, and are essential if a robot is to be able to move about a user's home effectively on the basis of a routine or in response to user commands. Robotic manipulators are available and continue to increase in sophistication, from robotic arms to assist people with dexterity impairments to complex manipulation devices able to fetch and carry, but this functionality escalates costs considerably (e.g. Abudulrazak, Mokhtari and Grandjean Reference Abudulrazak, Mokhtari and Grandjean2004; Graf, Parlitz and Hägele Reference Graf, Parlitz, Hägele and Jacko2009).

Much technical effort is currently focused on robots able to track and recognise people, interpret gestures and identify from a person's posture whether they are anxious or not feeling themselves (Cesta and Pecora Reference Cesta and Pecora2005; Hacque and Prassle Reference Haque and Prassle2010; Perrin et al. Reference Perrin, Chavarriaga, Colas, Siegwart and Millán2010). Several research groups are working on mobile service robots able to communicate verbally, undertake video monitoring, sense vital signs and deliver medicines, with examples including Kompaï, CareBot, CompanionAble, Florence Nightingale and Pearl (Badii et al. Reference Badii, Etxeberria, Huijnen, Maseda, Dittenberger, Hochgatterer, Thiemert and Rigaud2009; Besio, Caprino and Laudanna Reference Besio, Caprino and Laudanna2008; Dautenhahn et al. Reference Dautenhahn, Woods, Kaouri, Walters, Koay and Werry2005; Pollack et al. Reference Pollack, Engberg, Matthews, Thrun, Brown, Colbry, Orosz, Peintner, Ramakrishnan, Dunbar-Jacob, McCarthy, Montemerlo, Pineau and Roy2002). Work on human–robot interactions has identified the information needed by both humans and robots for different levels of interaction, and human–robot interaction standards have been derived and evaluated in tests (Goodrich and Schultz Reference Goodrich and Schultz2007; Hearst Reference Hearst1999; Scholtz Reference Scholtz2003; Tsui et al. Reference Tsui, Yanco, Kontak and Beliveau2008). As well as multimodal dialogue systems that can perceive people in their environment and both recognise and synthesise speech, three-dimensional gesture recognition has been developed from virtual reality games (e.g. Nickel and Stiefelhagen Reference Nickel and Stiefelhagen2007; Nieuwenhuisen, Stueckler and Behnke Reference Nieuwenhuisen, Stueckler and Behnke2010).

Automated understanding of the emotional or physiological state of a person is a current research frontier in robotics (Okamura, Matarić and Christensen Reference Okamura, Matarić and Christensen2010). The use of wearable devices to track vital signs such as heart rate is well-established in telehealth, but there is a range of situations when signs of something being wrong can be observed visually. Robots can undertake visual monitoring such as of facial expressions and some are already able to recognise signs of pain (Littlewort, Bartlett and Lee Reference Littlewort, Bartlett and Lee2009). A robot's analysis of natural language and non-verbal expressions can recognise different emotional states as a basis for interaction (e.g. Saint-Aimé et al. Reference Saint-Aime, Jost, Le Pévédic and Duhaut2010). Other advances, in the field of robot intelligence, include learning and predicting human behaviour. A robot can learn what to expect and respond with a verbal warning or an alarm to a third party if what it encounters is not what it anticipates, such as distress or risky behaviour. This can extend to monitoring the environment, detecting hazards such as dangerous obstacles, smoke or a gas escape. Further advances in evolutionary robotics are enabling robots to cope with situations without being programmed with prior knowledge of them, creating intelligent and autonomous robot controllers (Lee, Kimand Bien Reference Lee, Kim and Bien2010; Nelson, Barlow and Doitsidis Reference Nelson, Barlow and Doitsidis2009; Nolfi and Floreano Reference Nolfi and Floreano2004).

The RoboCare project has investigated the acceptability and usefulness of a socially interactive robot in the home, with positive findings especially for increasing personal safety (Cesta and Pecora Reference Cesta and Pecora2005). The CompanionAble robot is being developed for people with chronic cognitive impairment, embedding the robot in a smart home as part of a total assistive environment (www.companionable.net). The University of Hertfordshire's Robot House is experimenting with a robot in an ordinary home environment, exploring human–robot interactions regarding routine home tasks such as carrying food from a fridge, as well as investigating aspects of user acceptability (Koay et al. Reference Koay, Syrdal, Walters and Dautenhahn2009). User acceptance by older people in home settings has been a focus of several studies with generally positive findings, especially if the robot can communicate, but with concerns about capability limitations, reliability and the price of both the robot and its maintenance (Broadbent et al. Reference Broadbent, Kuo, Lee, Rabindran, Kerse, Stafford and MacDonald2009; Cesta et al. Reference Cesta, Cortellessa, Giuliani, Iocchi, Leone, Nardi, Pecora, Rasconi, Scopelliti and Tiberio2000; Heerink et al. Reference Heerink, Kröse, Evers and Wielinga2009).

Paro, a fur-covered robot baby seal, is one of the best know robot pets already on the market, specifically developed for older people to stroke and interact with, and costing around £4,000. These have been evaluated for their therapeutic benefits for people with dementia, with positive results for emotional state, reducing loneliness, and increasing communication and interaction (Banks, Willoughby and Banks Reference Banks, Willoughby and Banks2008; Kanamori, Suziki and Tanaka Reference Kanamori, Suziki and Tanaka2002; Song Reference Song2009; Tamura et al. Reference Tamura, Yonemitsu, Itoh, Oikawa, Kawakami, Higashi, Fujimooto and Nakajima2004). However, these devices when used by people with dementia or very young children raise particular ethical issues concerning the illusion of sentience, which is considered in the next section.

In summary, there is much useful technical development and evidence about acceptability, but this has so far not resulted in significant adoption of a product, even though robot technology can clearly deliver the benefits identified by evaluations of telecare and telehealth from vital signs monitoring, safety and security monitoring to information, advice and support. Crucially, an affordable product has yet to be developed based on market research. This consumer-focused approach implies giving as much – if not more – attention to attractive design, engaging functionality and intuitive usability as to technology development. In addition, although research in this field is increasingly multidisciplinary, the contribution of the social sciences is under-developed. An important example of where this is needed is seeing both the person with assistive needs and their carer/s in the context of their social networks. Any intervention needs to work with and strengthen these networks, which are essential to the psychological wellbeing of older people, and a robot needs to be seen as part of a human–machine network rather than a standalone device (Blackman, Brodhurst and Convery Reference Blackman, Brodhurst and Convery2001; Burnett et al. Reference Burnett, Regev, Pickens, Prati, Aung, Moore and Dyer2006; Coughlin, Pope and Leedle Reference Coughlin, Pope and Leedle2006; Duner and Nordstrom Reference Duner and Nordstrom2007; Eloranta, Routasalo and Arve Reference Eloranta, Routasalo and Arve2008; Golden et al. Reference Golden, Conroy, Bruce, Denihan, Greene, Kirby and Lawlor2009; Walker and Hiller Reference Walker and Hiller2007).

A care robot may have its own effect on loneliness given some existing evidence on companionship, but it also has a key role in creating supportive networks around the user (Banks, Willoughby and Banks Reference Banks, Willoughby and Banks2008; Kanamori, Suziki and Tanaka Reference Kanamori, Suziki and Tanaka2002; Wada, Shibata and Kawaguchi Reference Wada, Shibata and Kawaguchi2009). Robots can be linked to a mobile phone and web applications that provide a means for carers to keep in touch, receive and respond to messages, and programme the robot remotely. The robot is an actor in this networking, such as alerting a carer or circle of carers to an emergency or possible risk, or enabling a group of carers to purchase a robot jointly which is then used to share periods of respite and, via a related web application, network and share news and advice. This concept draws upon the idea behind www.justvisiting.com, a recent start-up social enterprise in North East England that facilitates a circle of friends and relatives keeping in touch with a hospital patient.

Ethical issues

There are several ethical concerns about assistive robots, mainly concerning the possibility of inappropriate attachment to them by users, privacy, the robot not distinguishing between lower- and higher-level risks, and the confidentiality of information collected by the robot, such as visual images. These have been most exhaustively explored in relation to child-care robots that are already on the market (Sharkey and Sharkey Reference Sharkey and Sharkey2010). These both monitor the child by transmitting images to a mobile phone or computer and occupy the child with activities. Concerns have been raised about addiction and emotional and psychological damage if children are exposed for large amounts of time to the robots at the cost of human interaction, especially with parents and carers. However, it is also argued that the risks are over-stated, with other authors drawing parallels with the pretend play typical of childhood, emphasising educational and therapeutic benefits, and citing evidence of how children can be very aware of the limitations of robots (Belpaeme and Morse Reference Belpaeme and Morse2010; Feil-Seifer and Matarić Reference Feil-Seifer and Matarić2010; Movellan Reference Movellan2010).

There is an argument to be made that design and functionality should not suggest a real animal, possibly implying having its own mental state and emotions, since this might be regarded as deception and engender inappropriate bonding, especially for people with cognitive impairments. It may also invite others to believe that their own time caring for someone or just keeping them company can be adequately substituted with a robot. A further risk from creating the illusion of sentience and emotions is that the user may dangerously exaggerate what the robot can do for them. Nevertheless, the Paro robotic baby seal is already being purchased by health and social care services as a social companion for people with dementia, and as already noted has been shown to encourage nursing home residents to interact with each other (Belpaeme and Morse Reference Belpaeme and Morse2010; Feil-Seifer and Matarić Reference Feil-Seifer and Matarić2010). It is also important to recognise evidence that how people perceive and respond to robots, and the relationships they form with them, will not be the same for everyone. These have been found to vary according to pre-existing attitudes and prior exposure to robots in literature or entertainment media, although not by age or sex (Broadbent et al. Reference Broadbent, Kuo, Lee, Rabindran, Kerse, Stafford and MacDonald2009).

For older people, the main issues are raising awareness about risks, hazards and appropriate use among both users and carers, which apply to any new technology including telecare. Suggestions that telecare is intrusive, reduces personal contact or causes isolation have not been supported by evaluations. Robots raise additional issues but the potential for harm can be minimised by describing as far as possible their capabilities and limitations to users and carers as guidelines, possibly underpinned by regulation (Whitby Reference Whitby2010). Ganyo, Dunn and Hope (Reference Ganyo, Dunn and Hope2011) suggest that manufacturers should be under a duty to inform potential buyers and users of the ethical issues raised by telecare products and the unwanted effects that may arise, making a parallel with how the pharmaceutical industry is required to alert patients to the possible side-effects of its products. However, this may continue to frame and indeed stigmatise telecare as special needs devices rather than consumer products. Marketed robots are already covered by the same consumer protection legislation as any product regarding instructions, warnings and a duty of care to the customer (Lichocki, Kahn and Billard Reference Lichocki, Kahn and Billard2011). If these responsibilities are appropriately enacted there does not seem to be a good argument for regarding care robots or other assistive technology as a special case for ethical protection.

The zoomorphism and anthropomorphism characterising much care robot development is probably unstoppable given that products are already on the market and views are divided about the ethical issues, reflecting both personal and cultural differences of opinion. However, aside from the ethical concerns, these features are not necessarily helpful in marketing care robots as really new products beyond the niche market of therapeutic pets that do not need to be fed or toilet trained. This is because a more obvious and less problematic source of comparison for consumers is not animals and humans but products that enable prior knowledge to be used to build representations of the really new product and what it can do as something that is both useful and desirable. These are other well-designed domestic appliances and personal technologies: assistive robots are a hybrid of a domestic appliance such as a vacuum cleaner and a personal technology such as a smart phone, and there is potentially a mass consumer market for them.

Care robots are telecare's ‘adjacent possible’

In 2010, the telecare device industry turnover in the UK was estimated at only about £100 million and fairly stable (Technology Strategy Board 2010). The market is still undeveloped and, as already noted, there is currently no significant consumer market. In the UK, uptake even by public services has been slower than expected, although is certain to be boosted by the WSD results and the Department of Health's commitment in England to extend telecare and telehealth to at least three million people with long-term conditions and/or social care needs (Department of Health 2011). It has been argued that to date this slower than expected pace of adoption is not so much due to limitations of the technology as to poorly joined-up service models, resistance to change by frontline staff, a lack of commissioning knowledge and a culture of thinking in terms of hours of care (Moore Reference Moore2011; Yeandle Reference Yeandle2009).

While it might be possible to improve uptake by public services using policy drivers, the potentially very large latent consumer market needs a different approach, with products that are not perceived as disability aids but are stylish and engaging as well as useful, integrate with the existing lifestyles of users and family carers, and are intuitive to use. This is also likely to mean that such products when procured by the public sector are more appealing to users, less stigmatising and cheaper because of the larger market.

The assistive robot concept offers a solution that is not aimed at people needing complex assistance but at the much larger numbers needing inclusive technologies with helpful features. In addition, while the product would be aimed at private purchasers, these will increasingly include people referred to public health and social care services in countries such as England where much direct provision by these services is being replaced with personal budgets enabling users to make individual purchasing decisions (Bönker, Hill and Marzanati Reference Bönker, Hill, Marzanati, Wollmann and Marcou2010).

The United States of America is likely to take a lead in marketing high-cost assistive robots starting at around US $12–15,000, with the price justified by a comparison with residential home costs (Owusu Reference Owusu2010). However, an alternative concept is to focus on meeting low- to moderate-level needs that may then help prevent escalation to higher-level needs, with entry-level pricing of a few hundred pounds and the option to spend more on specialist modules, either for purchase or hire. The product would be likely to substitute for many telecare packages as a more flexible and customer-friendly alternative, but more importantly also expand the market significantly. This in turn would have the potential of supporting the growth of a low-cost care robots sector.

Rather than a technological leap, such a development is a move into what Johnson (Reference Johnson2010) calls the ‘adjacent possible’ in innovation. Figuratively, this is the empty room next door to already occupied rooms where new innovations arise from novel combinations of existing technologies and/or the adaptation of an existing technology to a new use. From this perspective, affordable assistive robots are a special case of, on the one hand, the now common use of telecare and, on the other, small robots such as robotic vacuum cleaners. From a user viewpoint, a small robot that can perform functions similar to existing telecare equipment installed into a person's home promises a more flexible and portable option, as well as physical embodiment in one personal device, which some studies suggest older people prefer to mediation through a telephone or computer screen (Heerink et al. Reference Heerink, Kröse, Evers and Wielinga2009; Tapus Reference Tapus2009).

If such a product can be developed, it also promises to reframe the arguments often made for telecare that we face unsustainable costs from an ageing population. Rather than this ‘ageing burden’ narrative, assistive robots can appeal to lifestyle needs and successful ageing (Neven Reference Neven2010). Their negative association with dependency could also be addressed by having a wider market than just personal care, such as remote surveillance of an unoccupied home, and the portability of the robot means that it can be taken on holiday, used for only temporary spells or shared among a group of users and carers as respite. It is a product that should be wanted as well as needed.

More research and development is needed to keep the potential selling price low while incorporating important functionality but this is incremental rather than transformational R&D. Software architectures with search algorithms for finding objects and selected localisation are feasible, as are mapping and obstacle avoidance algorithms to provide navigation capability in cluttered and changing indoor environments. Development challenges and costs begin to escalate with providing mechanical and physical assistance, even though this offers potential for aiding manipulation, fetching and carrying, or opening doors. It is not a direction for a relatively low-cost device but could be available as add-on modules at extra cost depending on market demand.

More promising for a mass home market is that functionality embraces personalised communication, monitoring of the older person's state and their environment, and a range of response protocols from interaction with the individual user to communication across their social or family network. This last requirement is about future-proofing. Although the vast majority of telecare monitors are connected via a regular telephone line, increasing numbers of users and many carers (including family members and neighbours) have a mobile phone. An affordable care robot therefore needs to provide mobile phone connection and be ready for mobile phone video-conferencing services. Broadband internet connectivity needs to be built in, allowing remote communication and programming, as well as linkage with a website that also encourages and supports social networking. The use of social media platforms such as Facebook is set to spread even further with the growth of digital TV ownership and online access using familiar devices, as well as easy-to-use tablet computers, which are also likely to see improvements for users with impairments.

Rolling these functions into a single attractive design creates a really new product, although based on incremental technology development. The marketing literature, however, suggests great caution about bringing ‘really new products’ to the market, since they have a much higher failure rate than products that are ‘incrementally new’ mainly due to a lack of understanding of the product's features and benefits by consumers (Feiereisen, Wong and Broderick Reference Feiereisen, Wong and Broderick2008). Steps to address this have focused on using marketing communication strategies that draw on analogies and simulation of the potential purchaser in a consumption situation. Examples for a care robot include the robot vacuum cleaner, a conventional telecare alarm, smart phones and social networking sites such as Facebook. A simulation might be envisaged of an older person in their home being prompted by the robot about a meal time or querying why they are making for the front door in the middle of the night. Family members are as much, if not more, the target for these messages as the older person needing support.

As discussed above, the potential of robots to have ‘pet appeal’ has been demonstrated and, beyond a therapeutic value, helps to maintain the user's interest in the device and what it can do. At one level it is unavoidable that this type of attachment will occur with a small, attractively designed machine. Feil-Seifer and Matarić (Reference Feil-Seifer and Matarić2010) report how users of a robot vacuum cleaner become attached to it, including getting the device repaired when broken rather than replacing it. Although designing a care robot deliberately to create the illusion of sentience and emotions is questionable, there is every reason to design an affordable assistive robot in a way that is aesthetically pleasing as well as functional, not least to position it as a consumer product rather than an aid for special needs.

In reviewing the literature on assistive robot development for this article, there was a marked absence of commercialisation considerations, revealing a lack of marketing research input to the projects. Interestingly, a bibliographic search in Web of Knowledge combining ‘really new product*’ with ‘robot*’ produced no results, with the literature on robotics dominated by technological papers. Yet the research about really new products is extremely relevant and important insights are available, not least learning from product failures in technology markets due to poor customer acceptance. Important considerations in addition to price are targeting, positioning, and inter-firm partnerships and alliances critical to fostering adoption of the product. The last is especially important in technology markets because really new products are usually systemic innovations that are part of a wider system of interrelated components that need to respond with complementary actions, such as battery and software suppliers and retailers (Chakravorti Reference Chakravorti2003, Reference Chakravorti2004). Indeed, fruitful strategic partnerships between a robot developer and a business with large reach such as the ‘over fifties’ company Saga or a trusted brand such as Dyson could be envisaged.

Another important consideration regarding the mass market necessary to bring down costs for the consumer is diffusion. This can be significantly affected by post-purchase attitudes among early adopters, who can exercise considerable influence on the growth of the market for a really new product. Later adopters inevitably tend to be more risk averse, and relatively often early adopters develop a negative attitude towards the product because it fails to meet expectations (Chiesa and Frattini Reference Chiesa and Frattini2011). The lessons point to new innovations needing to arrive on the market completely developed and functioning perfectly. Configuration of the product on launch is therefore best confined to a limited number of functions that appeal strongly to early adopters, with this reflected in the product's advertising.