4Research, funding and innovation
75.We conclude this report by examining the research, funding and innovation landscape for robotics and AI and identifying barriers to progress.
Robotics and autonomous systems
76.In 2013, the Government identified ‘Robotics and autonomous systems’ (RAS) as one the its ‘Eight Great Technologies’, rather than using the ‘robotics and AI’ label, which is more commonly found in the digital sector. The ‘Eight Greats’ are technologies in which the Government anticipated that “the UK [was] set to be a global leader”.161 RAS was defined as:
interconnected, interactive, cognitive and physical tools, able to variously perceive their environments, reason about events, make or revise plans and control their actions. They perform useful tasks for us in the real world, extending our capabilities, increasing our productivity and reducing our risks.162
77.Though the different labels appear essentially to cover the same technology, witnesses did highlight that there was scope for confusion between ‘automated’ and ‘autonomous’ systems.163 Innovate UK noted that the vast majority of current robotics systems are automated, rather than autonomous, but that the concepts are sometimes used interchangeably, despite having different meanings.164 Both refer to processes that may be executed independently, from start to finish, without any human intervention. Automated processes, however, usually involve well-defined tasks that have known, consistent outcomes and are conducted in structured, predictable environments, such as factories. More often than not, an automated process replaces a routine, manual task with software/hardware that repeatedly follows a step-by-step sequence.
78.Industrial robots performing a limited range of “physically difficult, dangerous, or repetitive tasks” currently dominate this market.165 For example, ‘Latro’, a robotic spider can step around, or climb over, obstacles. Though Latro remains under development, it is anticipated that the robot will eventually be able to play an important role in UK nuclear decommissioning by grabbing waste, chopping it up and dropping it in a skip.166
79.An autonomous system, in contrast, will have the capability to learn, respond and adapt (within set boundaries) to situations that were not pre-programmed or anticipated in the design. This type of system may also be able to make decisions “based on external events and internal goals that lead to different courses of action, even when faced with unexpected events and unknown environments”.167 An autonomous car, for example, is designed to react to weather and traffic conditions. In this respect, there are similarities between autonomous systems and artificial intelligence. As robots become smaller, and more nimble, we are also beginning to see them working in conjunction with humans. The results to date have been encouraging, particularly in healthcare. Early in September 2016, a British surgeon restored a patient’s sight by operating a robot inside their eye, via a joystick, to “remove a membrane one hundredth of a millimetre thick”. The robot was described as “acting like a mechanical hand” that was “able to filter out hand tremors from the surgeon”.168
RAS 2020 Strategy
80.To stimulate collaboration and innovation in RAS capabilities, a RAS ‘Special Interest Group’ (SIG), comprising academics and industrialists, was established in 2013 with support from Innovate UK. The SIG subsequently produced RAS 2020, a national strategy for RAS, in July 2014. Its objective was “to capture value in a cross-sector UK RAS innovation pipeline through co-ordinated development of assets, challenges, clusters and skills” and included the following eight recommendations aimed at realising that goal:
Invest further in the five RAS strategy strands: coordination, assets, challenges, clusters and skills to build the UK’s RAS capability.
Establish the means for funding agencies to formally work together in execution, so that ideas, people and activity flow readily from basic investigation through early stage demonstration to fully trialled commercial product.
Establish a RAS Leadership Council to engage with senior leaders across a range of sectors in industry, academia and Government, providing independent advisory oversight of planning and execution of the strategy.
Further develop engagement with the EU, investors and corporate resources in the UK and overseas to fuel the development of the 5 strands.
Continue to consult widely on potential Assets and cross sector Grand Challenges.
Continue to develop dialogue with those involved in standards and regulation, such as BSI and CAA, to develop more detailed thinking.
Extend outreach and public engagement activities to continue changing public perceptions and improve understanding of public concerns.
Articulate to businesses and investors internationally (e.g. through UKTI) that the UK aims to be the best place to invest in taking RAS technologies to market.169
81.Drawing on analysis by McKinsey, the strategy estimated that by 2025 RAS technologies would “have an impact on global markets of between $1.9 and $6.4 trillion per annum”. Witnesses thought that the UK had the potential to take a lead in developing and adopting the next generation of advanced robotics, with Innovate UK singling out “the service segment, the global market for which is expected to grow from $7 billion in 2014 to $18 billion by 2020”.170 It concluded that a “sensible aspiration would be for the UK to capture perhaps 10% of this market”.171
82.At present, however, the take up of robotics in the UK is low when compared to other countries. Paul Mason, Director of Emerging and Enabling Technologies at Innovate UK, told us that “the numbers for last year [2015] show that installed shipments of robots in China were 75,000 compared with 2,400 in the UK”.172 Professor Philip Nelson of Research Councils UK thought that the UK was probably “off the pace […] in the deployment of robotics” and, consequently, was probably losing market share.173 He noted that South Korea had been investing “$100 million per year [in robotics] for the past 10 years or so”, while Japan had “just put $350 million into a big programme of assistive robotics”.174 The UK, in contrast, had been identified in a study by the Copenhagen Business School—which modelled how much productivity in manufacturing would increase if all industries in a country had the highest-found level of robot-intensity—as “having the greatest potential for improvement”.175
83.Our inquiry highlighted three barriers inhibiting the type of progress envisaged in the RAS 2020 strategy—funding, leadership and skills shortages.
RAS Funding
84.Several of our witnesses questioned whether the current level of funding for, and investment in, RAS was sufficient to improve the UK’s position in the field.176 Some, for example, indicated that European Union, rather than UK, funding was underpinning the country’s strong track record in robotics and AI research. Geoff Pegman of R U Robots Ltd thought that the UK would not have its current capabilities in both robotics and artificial intelligence “without the European Framework programmes”.177 Similarly, Professor Rose Luckin from the UCL Institute of Education told us that “the funding that the research community has taken advantage of to hold its position internationally [in artificial intelligence and education research] has all come from the European Union”.178 In 2015, the Prime Minister’s Council for Science and Technology calculated that 80% of funding for RAS research came from the EU and stressed that “if this funding stream was disrupted then RAS industry and research in the UK may suffer as a consequence”.179
85.Others noted that the UK had invested in RAS, with Research Councils UK reporting that “EPSRC funding in the robotics research area [had] increased threefold since 2010 and currently [stood] at £33.8M”.180 The EPSRC’s UK-RAS Network stated that this funding had:
resulted in the creation of eight dedicated centres/facilities across the country covering key areas of transport, healthcare, manufacturing, and unmanned systems to ensure that the UK will maintain its leading engineering and research capacity in RAS.181
86.The Government also drew our attention to £19 million provided to fund four ‘EPSRC Centres for Doctoral Training’ in Edinburgh, Bristol, Oxford and Loughborough.182 While welcoming these investments, Research Councils UK worried that there was still a lack of “UK-wide critical mass required to move from isolated pockets of excellence to the formation of a national research, training and innovation infrastructure required to enable cross-sector exploitation”.183 It concluded that there remained “a need for greater coordination and visibility in RAS across both academia and industry”.184
87.A similar point was made by ABB Robotics, a supplier of industrial robots. It told us that the RAS Special Interest Group (SIG) had “done a great deal of work with universities in identifying the technologies of the future and [had] invested heavily in research and development” but that the SIG’s focus on universities had come at the expense of considering the needs of industry. 185 According to Mike Wilson of ABB Robotics, the RAS 2020 strategy had been “driven largely by academic interests looking at the more advanced robotic technologies”.186 The company complained that by:
concentrating on advanced manufacturing, the [Special Interest] Group [had] received little input from industrial robotics and UK manufacturing businesses. As such, the Group has had very little impact on productivity.187
RAS Leadership
88.Witnesses suggested that coordination between academia and industry could be improved through establishing cross-cutting leadership in RAS. Professor Nick Jennings, representing the Royal Society’s Machine Learning Working Group, told us that “lots of excellent work” was “going on in bits and pieces around the country” but that “something that tries to bring those together and show leadership” was required.188
89.Establishing a ‘RAS Leadership Group’ had been recommended in RAS 2020, with the aim of engaging “with senior leaders across a range of sectors in industry, academia and Government, providing independent advisory oversight of planning and execution of the strategy”.189 In response to RAS 2020, the then Universities, Science and Cities Minister, Greg Clark MP, stated in March 2015 that the:
Government agrees with this recommendation. We will establish a Leadership Council in Robotics and Autonomous Systems. I have asked officials to implement this recommendation.190
90.However, Dr Rob Buckingham, one of the co-authors of RAS 2020, told us that setting up a Leadership Council “did not happen; there was a change of Government and a slight change of direction”.191 Professor Nelson explained that the Research Councils were “still trying to co-ordinate matters in the absence of a leadership council” adding that “something like [a leadership council] to interact with, or be a governing body for, a national initiative in this area […] is going to be very important”.192
91.Dr Buckingham suggested that there was scope for the Government to take a much more active role in advancing the RAS sector. The RAS Strategy, he explained, “was not paid for or requested by Government; it was a slightly external thing done under EPSRC, KTNs and Innovate UK and we need it to be formally adopted”.193 Paul Mason from Innovate UK confirmed that RAS 2020 was “not an official Government strategy”.194 On that basis, he thought it “would be worth looking at a more overarching strategy document that might bring about strong Government action in areas like regulation, standards or procurement”.195
92.Research Councils UK advocated a complementary approach, namely establishing a RAS national flagship institute “with the dual purpose of accelerating emerging technologies to commercialisation and inspiring new disruptive academic research with the potential to open up new markets”.196 In a similar vein, Innovate UK told us that establishing an institute (or a Catapult)197, would “provide a coherent national focal point for market-led RAS activity, presenting a visible and open front door to engage end users and international inward investment”.198
93.When we asked the Government why the RAS Leadership Council had not been established, it stated that while the:
Rt. Hon Greg Clark MP said that the coalition Government would establish a RAS Leadership Council […] things have moved on since then and a number of activities in this area have been introduced to support the development of RAS.199
These activities involved strengthening “coordination and oversight between departmental interests in Robotics and Autonomous Systems, including through a cross-government senior officials meeting initiated by Cabinet Office”, as well as appointing a “RAS lead” in the Department for Business, Energy and Industrial Strategy, to “coordinate the efforts within the Department and Arm’s Length Bodies”. The Government also added that it intended “to utilise the upcoming Industrial Strategy to drive the progression of RAS”.200
RAS Skills shortages
94.In our reports on Big Data, Digital Skills, and Satellites and Space, we highlighted that the UK is facing a digital skills crisis.201 Witnesses to our current inquiry emphasised that while the demand for expertise in RAS and AI was booming, the UK lacked sufficient numbers of qualified people in these fields.202 According to Professor Nick Jennings, the UK still did “not [have] enough people […] doing the basic computer science required”203, while Dr Buckingham stressed that there remained a “vital” need for “more people who are good at STEM”.204 The RAS Special Interest Group were concerned that the growth of the UK robotics industry was being:
restricted by the output of engineering degree programmes and by competing industries. Of critical concern is the significant shortage of systems engineers; those engineers able to understand the complexities of developing systems and systems of systems that robotics and artificial intelligence are built on.205
95.There appears, however, to be a shift underway in the subject and career choices made by university graduates. Professor Nelson emphasised that the EPSRC’s centres for doctoral training in RAS were “over-subscribed”, with Research Councils UK reporting that some centres were receiving “an average of 100 applicants per year for around 10 places”.206 Similarly, Professor Stephen Muggleton, representing the Association for the Advancement of Artificial Intelligence (AAAI), reflected that while the majority of Imperial College graduates had previously gone into “the financial industry after they got their undergraduate degree”, they were now “increasingly staying on to do postgraduate work because the big take-up is in the tech industry”. He added that “companies like DeepMind” were “hoovering up quality people”.207
96.Professor Muggleton’s example of London-based DeepMind is illustrative of how the UK has been at the forefront of establishing new, AI-driven companies, a number of which have subsequently been acquired by the world’s biggest technology firms. DeepMind, for example, was acquired by Google for a reported £400 million in 2014.208 Less than two years later, Microsoft bought ‘Swiftkey’, the makers of a predictive keyboard powered by AI, in a deal worth an estimated $250 million.209 ‘Magic Pony’, a company which uses AI techniques to understand the features of an image, was bought by Twitter in the summer of 2016 “for an undisclosed sum rumoured to be about $150 million”.210
97.An editorial in Nature, however, recently warned that this type of “industrial migration”, out of computer science departments and into AI start-ups, could leave universities “temporarily devoid of top talent” thereby reducing “the number of students that can be trained, especially at PhD level”. Nature also questioned whether this movement of people could “ultimately sway the field towards commercial endeavours at the expense of fundamental research”.211
98.Despite identifying Robotics and Autonomous Systems (RAS) as one of the ‘Eight Great Technologies’ in 2013, Government leadership has been noticeably lacking. There is no Government strategy for developing the skills, and securing the critical investment, that is needed to create future growth in robotics and AI. Nor is there any sign of the Government delivering on its promise to establish a ‘RAS Leadership Council’ to provide much needed coordination and direction. Without a Government strategy for the sector, the productivity gains that could be achieved through greater uptake of the technologies across the UK will remain unrealised.
99.The Government should, without further delay, establish a RAS Leadership Council, with membership drawn from across academia, industry and, crucially, the Government. The Leadership Council should work with the Government and the Research Councils to produce a Government-backed ‘National RAS Strategy’; one that clearly sets out the Government’s ambitions, and financial support, for this ‘great technology’. Founding a ‘National RAS Institute’, or Catapult, should be part of the strategy.
161 HM Government, Eight Great Technologies infographic, October 2013
162 Special Interest Group, Robotics and Autonomous Systems, RAS 2020 Robotics and Autonomous Systems, July 2014, p 4
166 Nuclear Decommissioning Authority, The role of robotics in nuclear decommissioning, June 2016
167 Special Interest Group, Robotics and Autonomous Systems, RAS 2020 Robotics and Autonomous Systems, July 2014, p 7
168 “Robot operates inside eye in world first”, BBC News Online, 9 September 2016
169 Special Interest Group, Robotics and Autonomous Systems, RAS 2020 Robotics and Autonomous Systems, July 2014, p 7
172 Q86 [Paul Mason]
173 Q23; Q12
174 Q12
175 Q23 - the study estimated that productivity would rise in the UK by 22%. See Lene Kromann, Jan Rose Skaksen, Anders Sorensen, Automation, labor productivity and employment - a cross country comparison, CEBR, Copenhagen Business School, Working Paper, 2011
176 Manufacturing Technology Centre (ROB0018), para 14; Professor Tony J. Prescott (ROB0020); Royal Academy of Engineering (ROB0042) paras 22–23
178 Q87 [Professor Luckin]
179 Prime Minister’s Council for Science and Technology, Science Landscape Seminar Reports: Robotics and Autonomous Systems (RAS), June 2015
180 Research Councils UK (ROB0033) para 19. The EPSRC is the Engineering and Physical Sciences Research Council.
182 Department for Business, Innovation and Skills (BIS) (ROB0066) para 10; Department for Business, Energy and Industrial Strategy (ROB0076)
186 Q145
188 Q28 [Professor Jennings]
189 Special Interest Group, Robotics and Autonomous Systems, RAS 2020 Robotics and Autonomous Systems, July 2014, p 3
190 Minister for Universities, Science and Cities, Response to the Robotics and Autonomous Systems Strategy, March 2015
191 Q26
192 Q28 [Professor Nelson]
193 Q27
194 Q103
195 Q103
197 Catapult Centres are physical centres, established by Innovate UK, to “transform the UK’s capability for innovation in specific areas” through enabling businesses, scientists and engineers to work side by side. See http://www.catapult.org.uk/about-us/
198 Innovate UK (ROB0060) para 34; see also Research Councils UK (ROB0033) paras 38–40; Robotics & Autonomous Systems Special Interest Group (ROB0027) para 33; Northern Robotics Network (ROB0040) para 1.7
201 Science and Technology Committee, Second Report of Session 2016–17, Digital skills crisis, HC 270; Science and Technology Committee, Third Report of Session 2016–17, Satellites and space, HC 160, para 48; Science and Technology Committee, Fourth Report of Session 2015–16, The big data dilemma, HC 468, para 27
203 Q42
204 Q45 [Dr Buckingham]
207 Q41
208 Google buys UK artificial intelligence startup Deepmind for £400m, The Guardian, 27 January 2014
209 Microsoft buys British keyboard apps firm SwiftKey, The Guardian, 3 February 2016
210 Q105
211 Elizabeth Gibney, “AI firms lure academics”, Nature, vol 532 (2016) p 422
© Parliamentary copyright 2015
5 October 2016