Engineering and Physical Sciences Research Council Theme ... · The Theme Day was concerned with projects of relevance to robotics and autonomous systems and did not assess the full

Engineering and Physical Sciences Research Council

Theme Day Panel Report& Council Response

A NOTE ON TERMINOLOGYThis report concerns itself with an evaluation of the Engineering and Physical Sciences Research Council’s (EPSRC) portfolio of projects relevant to Robotics and Autonomous Systems (RAS); such systems are those that can gain information about their environment, learn, adapt and make decisions without the need for human control. It does not focus on the broader area of Robotics and Artificial Intelligence (RAI).

Professor David Hogg (Chair) University of Leeds

Dr. Mike Aldred Dyson

Professor Aude Billard École Polytechnique Fédérale de Lausanne

Dr. David Bisset iTechnic

Professor Henrik Christensen UC San Diego

Professor David Parkes Harvard University

Mr. Geoff Pegman RU Robots

Dr. Elizabeth Sklar King’s College London

Professor Sarah Spurgeon University College London

Panel members

Kedar PandyaEPSRC’s Associate Director for Business Engagement

This Theme Day report represents an independent assessment of the quality, impact and importance of EPSRC’s Robotics and Autonomous Systems (RAS) portfolio. A number of factors influenced my decision to seek an independent assessment of EPSRC’s RAS portfolio at this time. Chief amongst those is a recognition of the importance of continuing to stay at the leading-edge of thinking in supporting research and training in this fast moving field, within and across disciplines.

Another key factor is the critical importance of excellent collaborative academic research as a spring-board for the type of sustainable economic growth, skills and citizens’ well-being envisaged in the Government’s Industrial Strategy Green paper. As the UK holds its second national robotics week, as the Government plans for a long-term sustainable industrial plan and as we prepare for the formation of UK Research and Innovation (UKRI), the release of this report is excellently timed to galvanise the research community to meet the challenges that await.

The assessment is based on a representative sample of the RAS portfolio, with projects chosen for their relevance to robotics or the development of autonomous systems. Yet it cannot be ignored that the debate has moved forward, even in the relatively short period over which the review was scoped and delivered. For instance, there is now greater policy attention on the fundamentals of Artificial Intelligence and the potential for disruption in many fields, with other significant reviews now underway. While there are some connections to the RAS portfolio, in the main the EPSRC AI portfolio was not reviewed within this exercise.

I am immensely grateful to Professor David Hogg for his chairmanship of the panel. I am equally grateful to all the other panel members, drawn from international research establishments, UK industry and UK universities, for their contributions and their many insightful comments. Finally, and as always, I thank the EPSRC team who have demonstrated professionalism in both portfolio and knowledge management in delivering this report.

Professor David HoggPanel Chair, University of Leeds

This is an important time for the development of robotics research in UK universities and industry. The ground-work and preparation for this report has therefore been very rewarding. The recommendations of the panel are based largely on the Theme Day, and on behalf of the panel I’d like to thank everyone who took part for their enthusiastic and positive engagement, and for the excellent quality of the posters.

We were impressed by the quality of the work that we saw on the Theme Day and there was a clear consensus between panel members on the general conclusions and recommendations of our study.

The panel quickly established a strong working relationship and this was maintained throughout. My thanks go to them for their wise counsel on the content of this report and on the recommendations. Finally, I’d like to thank the key people involved at EPSRC for their excellent support of the panel, and particularly in the organization leading up to the Theme Day.

Foreword

IntroductionOn 31 January 2017, an independent panel carried out an evaluation of the Engineering and Physical Sciences Research Council’s (EPSRC) portfolio of research in the area of Robotics and Autonomous Systems (RAS)1. Named as one of the Eight Great Technologies by the UK coalition government in 2012, RAS was identified as a technology where the UK can be a global leader, revolutionising the economy and society over the next 20 years.

This is an area of high priority for the EPSRC and many of its research areas are expected to contribute to future developments in RAS. With RAS technologies increasing in prominence, a number of reports2 have highlighted the application areas where they have the potential to create economic and social gains. These areas include the following: • Extreme and Challenging (Hazardous)

Environments: RAS technologies allow for inspection, monitoring, and maintenance in areas that are dangerous for humans to carry out activities; for example, energy systems (e.g. oil and gas, nuclear plants, off-shore renewables) and infrastructure (e.g. bridges, roads, and rail).

• Health and Social Care: New surgical tools offer a greater degree of precision and accuracy, and technological developments will allow for the care of an ageing population at home.

• Autonomous Transport: The technologies that will allow for the autonomous transport of passengers and goods have the potential to transform a number of sectors; for example, enabling vehicles to carry out tasks in locations such as warehouses and hospitals without human guidance, and to conduct environmental monitoring on land, sea, and air.

• Next Generation Manufacturing: Robotics and automation have already revolutionised the manufacturing process in many industries and we are now on the brink of Industry 4.03. Aspects of this, such as new autonomous systems for manufacturing and cooperative behaviour between humans and robots, are important for the UK’s future competitiveness.

• Enabling Technologies: The ‘sector agnostic’, underpinning technologies that are fundamental to developments in RAS, such as developments in sensor technology, allow for the creation of a technological foundation that de-risks subsequent investment in industry specific applications.

These areas are of strategic interest to EPSRC and have been used to sub-divide the grants considered as part of this portfolio review. Through these illustrations, we see some of the ways in which RAS technology has the capacity to replace, augment or complement human activity, and has the potential to work collaboratively with humans or machines to significantly advance both physical and intelligent capabilities.

There is a need for additional evidence to support EPSRC understanding of the broad UK research landscape and to inform the development of EPSRC strategies in RAS and related areas. International benchmarking by an independent review panel was organised to provide this evidence and increase EPSRC understanding of opportunities and gaps within the research landscape.

1 Government strategy is currently referring to Robotics and Artificial Intelligence (RAS).

2 For example, RAS2020: Robotics and Autonomous Systems Strategy and The UK Landscape for Robotics and Autonomous Systems.

3 http://www.platform-i40.de

An EPSRC Theme Day is an evaluation mechanism where a panel of experts directly interact with investigators on peer-reviewed grants awarded in a specific subject area, either through poster sessions or seminars. The panel provides their opinions and perceptions on a representative sample of projects, which have previously been through formal peer review, from across the portfolio and draw conclusions about the area as a whole. Notably, a Theme Day is not concerned with constructing league tables of grants or researchers, nor to isolate individual successes or failures. In general, the purpose is to ascertain the effectiveness of EPSRC’s support to the area and the value of the scientific research in the area to the UK.

• Objective: A Theme Day helps EPSRC to understand the state of the portfolio of research activity in specific or interconnected areas. It also benchmarks the UK’s position against the international landscape for the research area concerned and assesses the quality and importance of the research area to relevant sectors and the UK economy. The Theme Day can also help raise the profile of the subject area.

What is a Theme Day?• Outcome: The evidence gathered as a result

of the Theme Day helps the EPSRC make a robust case for investment/intervention in the research area, and will inform as to the UK’s scientific strengths in the portfolio. In turn, this will allow EPSRC to facilitate advocacy for research in general as well as the particular research area.

• Output: The output is an independent report written by the panel highlighting strengths, weaknesses, opportunities, and threats to the UK’s activities in the research area, as well as an overall assessment of quality and recommendations for the future.

The Theme Day was concerned with projects of relevance to robotics and autonomous systems and did not assess the full spectrum of Artificial Intelligence research. Descriptions of how projects were selected for inclusion are on the next page. The recommendations are principally for EPSRC, although in some cases they are also intended for the wider RAS community and other stakeholders. This should be clear from the context and wording of each recommendation.

A list of funded grants with relevance to RAS was used as a starting point for selecting participants for the Theme Day. All grants of relevance that had been active or funded in the period April 2012 to October 2016 were initially included.

The portfolio managers with relevant expertise were then asked to review the list highlighting any proposals that were thought to have insufficient relevance for the purpose of the Theme Day, and adding any that were missing. This list was also reviewed with the panel chair, Professor David Hogg.

As well as identifying grants that will give a fair and honest representation of the RAS portfolio at EPSRC, it was also necessary to consider what would be a manageable number of grants for the panel to review within the Theme Day. All grants that have only been active since January 2016 were excluded from the process as it was judged that their outputs would be limited at this stage. Although First Grants form an important part of EPSRC’s RAS portfolio, they were not included due to space and time considerations. Capital awards were also excluded to focus on research and training.

The panel was asked to use the following criteria in evaluating the research and fellowship projects and the portfolio as a whole. This information was supplied to the grant holders in advance of the event to inform their poster presentation.

1. Research Quality: This encompasses a number of factors, including the novelty and timeliness of the project; the ambition, adventure, and transformative aspects identified; and the appropriateness of the proposed methodology. This should be assessed in an international context.

2. Impact: Academic / Economic and Societal: Academic impact of the projects will be evidenced with refereed journal and conference publications / presentations; book chapters; citations; invited keynote speeches. Economic and social impact will be evidenced by examples including dissemination, IP, spin out companies, industrial partners, training courses, etc. The panel was asked to consider whether the projects were making sufficient effort with regards to impact activities.

3. National Importance: The purpose of national importance is to encourage applicants to articulate why it’s important for their research to be supported by the UK taxpayer so that the UK remains internationally competitive. National importance has a number of strands and so answers to this question might cover:

• Why the research might benefit the UK economy?

• Why it may lead to advances in a different academic discipline?

• Why it’s important that an internationally leading group continues to be supported?

Project selection

Findings of the panel

RESEARCH QUALITY

1. The portfolio contains a broad spectrum of areas in robotics, including micro-manipulation, human-robot interaction, mobility, navigation, modelling, robot ethics, cognitive robotics, and bio-robotics. The panel was briefed about interesting and strong research in areas that reflect traditional industrial or computer science strengths in the UK, including autonomy for unmanned vehicles, autonomy for manufacturing technology, and formal methods (verification and validation) for robust and trustworthy AI systems. We were also pleased to see multidisciplinary collaborations involving social psychologists and computer scientists, investigating AI in the social context. The overall portfolio contains a considerable body of research that is internationally leading.

2. Work in some areas of RAS, such as robot learning, manipulation and grasping, compliant control, physical human-robot interaction and computational neuroscience were not prominently represented in the portfolio present on the Theme Day and may be gaps that need to be addressed. However, it is recognised that there are internationally recognised teams supported by EU funding and recently funded EPSRC projects in some of these areas.

3. The UK is present in a number of European projects, which will complete over the next four years. This is an important funding stream and there will be an impact if it is removed. Collaboration with European colleagues makes an important contribution to the research landscape, and consideration needs to be given to how groups identified as excellent in European terms can see their work continue, possibly even collaboratively with European partners, so that expertise is not lost.

4. It will be important for EPSRC to monitor the whole RAS research landscape over the next five years to ensure it remains balanced in line with national priorities and maintains critical mass of human capital.

RECOMMENDATION 1: EPSRC should work with the RAS community to identify critical gaps in investment across the whole UK RAS research landscape.

5. The fields of robotics, machine learning, and computer vision are going to see a strong confluence going forward. There is a need to find a balance between disciplinary and interdisciplinary research to ensure support for both. Thus, the field of robotics in the UK would benefit from greater collaboration with overlapping communities in machine learning and computer vision that are globally recognized and attractive to inward investors.

6. In general, matching best with best across the different communities that contribute to RAS would help to raise the proportion of research that is internationally leading and the overall impact of the research portfolio.

RECOMMENDATION 2: Encourage

research that links the strongest teams across the different fields that contribute to RAS.

7. There is potential for highly productive collaboration with disciplines outside computer science (e.g. materials, control). The axis between robotics and materials is growing rapidly in the US, and materials is an area in which the UK is already outstanding and investing for the future (e.g. through the creation of the Sir Henry Royce Institute).

8. There is a need for connecting activities between the RAS and AI communities. The gearing effect of this is likely to be significant. The indication from the portfolio review is that the connectivity is excellent in some places but can be improved overall. The Research Councils UK (RCUK) consideration of robotics and AI as a whole is a positive move in this direction.

9. There is a need to better balance the use of EPSRC resources between direct funding for research and ‘indirect’ funding for networks and community engagement. Lateral connections across technical communities are most likely to yield innovation and advantage. Consideration should be given to the development of inter-disciplinary and inter-technology coordination actions (as are used effectively within Horizon 2020) where strategic engagement priorities are developed and acted on rather than strategic technical priorities alone.

RECOMMENDATION 3: Consider the

development of inter-disciplinary and inter-technology connections, for example, promoting the interaction between robotics, machine learning and computer vision communities.

10. We saw several unrealised opportunities for collaboration between existing projects in the EPSRC portfolio. It could be beneficial to consider mechanisms that would encourage and facilitate ‘teaming’ of awarded grants and raise awareness of projects between each other and with the wider community. For example, a Theme Day devoted to Principal Investigators (PIs), with snapshot presentations/videos, or a website on which projects could promote their work, would help build the necessary connectivity and community.

11. Greater consideration needs to be given to the stimulation of collaborative projects that are multi-institution. There is benefit in human capital development and in technical collaboration that can be developed in key areas of activity to build projects at scale at the upper EPSRC-eligible Technology Readiness Levels (TRLs).

RECOMMENDATION 4: Consideration should be given to promoting collaboration between EPSRC-funded projects, and to stimulating larger-scale projects at higher Technology Readiness Levels involving multiple institutions.

12. State of the art experimental facilities and infrastructure are vitally important for research in robotics. Setting up major research infrastructure in robotics is expensive as there is a need not only for basic hardware but also for associated experimental facilities, including personnel who are highly trained to work with specialised pieces of equipment and other recurring costs such as maintenance. It would be timely to consider support for research infrastructure that could be shared across the community rather than having medium-scale facilities at multiple institutions. For closer to market research, the possibility of industry providing some of these experimental facilities should be investigated, perhaps linked with encouragement towards open innovation collaborations to meet key industrial long-term challenges.

13. Where personnel trained in operating specialised facilities are located at different sites, there is scope to enhance and transfer skills through encouraging an active national community across this group, including the possibility of short-term visits between sites.

RECOMMENDATION 5: Consideration should be given to developing shared infrastructure for research in RAS, including the potential for industry to provide experimental facilities relevant to key sectors and challenges.

14. Knowledge about RAS advances through an international effort, which is largely mediated by published outputs and conferences. Closer collaboration between leading research groups in different countries can be highly beneficial in accelerating the rate of discovery and the expertise within these groups. Joint funding is important for these closer collaborations to flourish, requiring either coordination between national agencies (e.g. between EPSRC and the National Science Foundation), or transnational funding agencies (e.g. EU).

15. With the risk of a reduction in funding for UK institutions from the EU and to ensure the current connections between the UK and Europe do not reduce over time, it

would be expedient and timely to address European national to national research links as early as possible.

RECOMMENDATION 6: The RAS research community and EPSRC should work to sustain and develop deeper international research links and joint funding opportunities, both within Europe and beyond.

IMPACT

Academic

16. A major part of the RAS research undertaken with EPSRC funding is appearing in the leading archival conferences and journals in the field. However, the extent of publication in leading journals, such as the International Journal of Robotics Research, and IEEE Transactions on Robotics, does not seem to have been given appropriate attention. Publication in high quality journals remains important for significant contributions because they reach additional audiences, encourage a more thorough exposition of results, and support promotion and tenure. There is therefore scope to increase academic impact further by targeting high-quality journals to a greater extent than is currently the case.

17. The overall research portfolio visible on the Theme Day appeared less focused on fundamental methods and algorithms than research funded by other bodies such as the US National Science Foundation. Whilst it is recognised that such work may not have been visible in the sampling of projects viewed on the day, it is important, in protecting the UK’s long-term interests, that the EPSRC continues to have a balanced portfolio of projects that includes both longer-term curiosity driven research and research that is closer to societal needs and industry challenges.

Economic and Societal

18. In most cases, researchers funded within the current RAS portfolio are engaging closely with industry in setting their research agenda. There were good examples of large industry engaging

proactively with universities in the pursuit of longer-term application-oriented goals. In this respect, it is important to have SMART (Specific, Measureable, Attainable, Realistic, and Timely) key performance indicators that measure translation and impact. Engagement is critical as a precursor to translation, and thus worthy to measure, but it is translation that matters.

19. Some of the more fundamental research in the portfolio is narrow in scope with little obvious chance of impact. In other cases, where there is excellent foundational work being done, there is also an opportunity for innovative application, but without specific plans to do this being in place. The portfolio contained some excellent examples of needs-driven research but also examples of research where too little thought had been given to possible paths to impact.

20. While curiosity-driven responsive mode funding is essential in many areas of novel and early-stage research, and fundamental research should be supported, much of RAS has direct application. There is a role for early stage risk reduction where excellent science is coupled to disruptive innovation addressing application need, in essence, addressing areas where “market failure” in top down research investment is occurring. Novel techniques often encounter significant market barriers, and early engagement from commercial organisations can highlight and mitigate these early stage risks, accelerating both investment and deployment. Excellent innovation is not incompatible with excellent science.

21. As foundational research moves into technology development, it is essential to promote engagement with the market, as is a collaborative approach even at low TRL levels. Novel technology cannot be “landed” in a market; it must be eased in gently. Technology should not be developed in isolation from a real market need.

22. Thus, whilst there is a need for greater emphasis on cutting-edge fundamental research, there is also the need and

opportunity for a greater proportion of the overall portfolio to be linked to societal needs and industry challenges, particularly for research that is more applied in nature. This linkage can be stimulated through greater industrial engagement, as seen in the ‘Towards Autonomy – Smart and Connected Control’ (TASCC) initiative, in the previous ‘Autonomous Intelligent Systems Partnership’ (AISP), and in the Oxford autonomous vehicles projects.

23. Indirect support for Principal Investigators in linking foundational research with innovative application aligned to industrial and societal needs should be balanced with direct support through funding. This indirect support can happen at an individual project level with project advisory boards that should be capable of delivering the national strategic picture at a “project local” level, and also through national coordinating events. In addition, active development of engagement with research strategies in RAS should also be supported. RAS is a diverse field and its national sector impact is potentially very broad.

RECOMMENDATION 7: Ensure greater emphasis in the RAS portfolio on excellent foundational work, with a larger proportion of the wider portfolio linked to industrial and societal needs and opportunities, and positioned to translate into innovative application.

24. There are outstanding examples of ‘innovation engines’ driven by individual researchers from within the UK RAS community. Consideration should be given to new mechanisms or existing schemes that can empower and motivate other leading researchers to adapt and replicate these successes. For example, allowing a period of time in which a researcher can focus on building a team, its competencies and infrastructure, without the distraction of having to devote time and effort to raising funds, enables a project to start on a strong footing from which innovation can thrive.

RECOMMENDATION 8: Seek ways to replicate the successful innovation mechanisms that exist in parts of the RAS research community.

25. With some notable exceptions (e.g. for autonomous vehicles), there is a need for more field-testing to demonstrate the viability of new methods in environments that are closer to an applied setting. This is important for industry to be able to evaluate new technologies.

26. Consideration should be given to support for national testing facilities targeted at the upper end of EPSRCs TRL remit, with resources allocated to providing open access and facilitation in addition to direct research funding. Synchronisation is essential not only with Innovate UK but also with other research councils and national research organisations such as the National Oceanography Centre and the National Nuclear Laboratory. The provision of such facilities should be coordinated with a wider need for shared experimental facilities and infrastructure.

RECOMMENDATION 9: Provide national testing facilities to demonstrate the viability of new methods and to promote take-up by industry.

27. There is good engagement from the research community with regulatory and approvals authorities, which is an essential link for maximizing economic and societal impact. Still, consideration should be given as to how access to these bodies can be made easier and how to promote the development of regulation and accreditation where it is immature or non-existent. EPSRC could work with organisations such as the Lloyd’s Register Foundation in achieving this.

RECOMMENDATION 10: Promote the development of regulation and accreditation of new RAS technologies to maximise economic and societal impact.

28. There is new potential for robotics in relation to the growing commercial and academic interest in the Internet of Things and Big Data. Interdisciplinary work in this area will enhance the UK’s research credibility. Given the foreseen growth in networked robotics and the use of robotics as the business end of the Internet of Things, work in this area will underpin some of the significant market growth areas forecast for RAS.

29. There is increasing demand for RAS researchers across academia and industry that shows no sign of abating. It will be important that a growing number of researchers reach the level of knowledge and expertise provided by doctoral training. Some collaborative activity conducted through the Centres for Doctoral Training or in conjunction with bodies such as the professional engineering institutions could be used to create impact earlier in the educational cycle to ensure the opportunities and demand for RAS are communicated by exemplars.

NATIONAL IMPORTANCE

30. The UK not only has a strong basic research program in RAS, but has also recently seen the emergence of a strong UK RAS research presence in several industry sectors, particularly in aerospace, automotive, and information technology.

31. There is potential for the UK to broaden, and in some cases rebalance, the RAS research engagement within a number of other sectors, including agriculture, construction, entertainment, logistics/warehousing and Unmanned Aerial Vehicle (UAV) technologies. It is not evident there is a structured collaboration with basic research and the downstream mechanisms for commercialisation associated with these sectors. Plans for greater integration across the research councils and Innovate UK will help in this regard.

32. The scale of the NHS is a unique asset for the UK in the development of RAS for human health. The opportunity to improve public health through RAS is extensive. National centres exist but require national strategic cohesion and exchange. Pre-competitive Procurement

and the Public Procurement of Innovation type mechanisms fit well within the UK healthcare sector. While these may sit within the Innovate UK remit, there is a need for continuity of purpose between EPSRC and Innovate UK driven by NHS needs. NHS-EPSRC collaboration has the potential to create advanced medical practices across both intervention, care and logistics operations across the NHS. These practices cannot be developed by technologists alone and require advanced technical research combined with clinical and healthcare practice expertise. If clinical trials are needed, it takes a decade from an excellent idea to make this a product for the NHS because of the time to conduct the trials and obtain medical certification. Funding and support for such interdisciplinary endeavour therefore needs to be ensured over long periods of time by institutions and funding agencies.

33. Other areas of significant Public Procurement should also not be overlooked as potential sources for boosting the UK’s competitive advantage in RAS. Areas such as defence, transport and nuclear decommissioning could be harnessed if a strategic research and industry delivery mechanism could be put together between EPSRC, Innovate UK and the relevant government departments. The recent announcement of a major grant for Robotics for Nuclear Environments goes some way towards such aspirations.

RECOMMENDATION 11: Ensure there is continuity of purpose between EPSRC, Innovate UK and relevant government departments for the development of RAS within key industry sectors.

34. It is important for the UK to remain as a key collaborator with major industrial users and developers of RAS technologies. Amongst other things, this will ensure UK researchers continue to develop access to non-public domain datasets. Such access promotes collaboration and accelerates research progress, where consistent with commercial objectives. Critical to this is an identification of centres

of excellence in RAS in the UK and an assessment of their international profile. The Research Excellence Framework (REF) might provide the underlying data but this is typically historic, which in a fast-moving field such as RAS may not deliver enough reactive capacity.

RECOMMENDATION 12: Maintain an up-to-date profile of international excellence and specialisation in RAS across UK institutions.

35. The UK already benefits from large, textured data in the public domain, for example anonymised electronic medical records. But beyond this, as companies such as Deep Mind and Amazon advance research in reinforcement learning, AI and autonomous vehicles, the EPSRC should form strategic alliances to ensure that these companies provide public data sets, or protected data sets available expressly for research purposes, that can push forward research within academia. This is important in sustaining a strong pipeline of STEM researchers, giving university researchers the ability to work at the frontier of problems that are representative of major challenges in scaling up AI and robotic technology. Industry also stands to benefit by the co-production of the new breakthrough ideas that can be translated back into practice and implemented at scale.

RECOMMENDATION 13: Seek ways to drive innovation in academia through bringing new datasets into the public domain from major information-centric companies.

36. With the UK leaving the EU, there is the potential loss of a route to market for research that hitherto has been carried out in collaboration with EU robotics companies.

TRAINING

37. The performance of Centres for Doctoral Training (CDTs) looks positive, with joint funding and collaboration with industry happening as planned. The first cohorts

are now well into their periods of study but have not yet graduated. It is too early therefore to evaluate their destinations post-PhD. However, it will be important to put in place a mechanism to keep track of alumni from the CDTs to both assess the quality of the CDTs and to build a network of RAS graduates working in the industrial and academic sectors. Some CDTs overlap in part in the topics covered. This is unavoidable. While it may lead to a healthy competition, one must ensure that each CDT conserves a unique contribution.

RECOMMENDATION 14: Ensure there is a mechanism in place to evaluate the early career pathways of alumni from the CDTs.

38. It will be important to ensure that industry continues to invest in CDTs in the future, renewing commitments that have been made and attracting new industry partners. Effort should be made to assess what has previously been successful in attracting investment and to ensure that this is maintained, as well as identifying what more could be done to promote CDTs to a wider industrial audience.

RECOMMENDATION 15: Review the success factors driving the participation of industry in the CDTs and distil practice for the future.

39. The PhD research projects across CDTs appear to be predominantly on the software aspects of RAS, with relatively few devoted to hardware innovations. This may not be the optimal balance for the UK as a whole, and some consideration should be given to increasing the research focus on hardware.

RECOMMENDATION 16: Review the balance between software and hardware research across all CDT PhD topics to inform the future direction of any potential RAS CDT programme.

40. Gender diversity is a very significant concern, with reported under-representation of females in the current

cohorts despite initiatives to redress the balance by individual CDTs and through the UK-RAS Network’s School Robot Challenge. More needs to be done to promote gender diversity within the CDTs that are supported as part of the RAS portfolio.

41. In regard to improving the pipeline of applicants, CDTs should work collaboratively to develop and share best practice. Outreach activities should be promoted, for example “Robotics Challenge” style events that bring students together to work on well-motivated, creative problems, while learning about modern AI and robotics research.

42. The EPSRC may find it beneficial to do public outreach, for example in regard to the modern AI and robotics workplace, as exemplified by companies such as Google/Deep Mind and Amazon, as well as to the critical importance of AI and robotics in maintaining the competitiveness of the UK manufacturing sector.

43. There should be support for the women in CDTs across the UK to come together, in ways similar to the Grace Hopper Celebration of Women in Computing in the US. Research has shown how important it is for people to have peers with whom they can relate as they advance in academic careers.

44. The EPSRC should mandate a regular survey of doctoral students in all current CDTs, looking to understand differences (if any) in the experience by men and women in current programs and to track progress.

RECOMMENDATION 17: CDTs that are part of the RAS portfolio should develop and share best practice on promoting gender diversity and collaborate in outreach activities to maximise impact. EPSRC should put in place a survey mechanism.

45. Each CDT operates an independent admissions process. It would be beneficial to applicants for there to be a mechanism by which they could be directed to the CDT that is most appropriate for their interests, highlighting the specialisms of each CDT. The CDTs could directly collaborate with

one another to provide such a mechanism in a way that makes clear where they are complementary to each other and where they are distinctive. Collaboration in this way would also help in mapping out and planning the landscape for a potential future CDT funding activity.

46. There is further scope to create a national community of RAS PhD students from the different CDTs. One mechanism for ensuring this could be through the organisation of summer schools and related initiatives such as joint research competitions. The UK-RAS Network could provide the stimulus for this.

47. Coherency of process and propagation of best practice would help in promoting transfer and short-term exchange between CDTs.

RECOMMENDATION 18: CDTs should consider how they could work together and with EPSRC to (i) signpost the RAS CDT portfolio to potential applicants, (ii) develop a national community of RAS PhD students, and (iii) harmonise and propagate best practice to promote mobility between CDTs.

48. From a national perspective, it is important to ensure graduate and technical training in RAS is considered alongside the CDT programme.

Summary of recommendations

EPSRC should work with the RAS community to identify critical gaps in investment across the whole UK RAS research landscape. Encourage research that links the strongest teams across the different fields that contribute to RAS. Consider the development of inter-disciplinary and inter-technology connections, for example, promoting the interaction between robotics, machine learning and computer vision communities.

Consideration should be given to promoting collaboration between EPSRC-funded projects, and to stimulating larger-scale projects at higher Technology Readiness Levels involving multiple institutions.

Consideration should be given to developing shared infrastructure for research in RAS, including the potential for industry to provide experimental facilities relevant to key sectors and challenges.

The RAS research community and EPSRC should work to sustain and develop deeper international research links and joint funding opportunities, both within Europe and beyond.

Ensure greater emphasis in the RAS portfolio on excellent foundational work, with a larger proportion of the wider portfolio linked to industrial and societal needs and opportunities, and positioned to translate into innovative application.

Seek ways to replicate the successful innovation mechanisms that exist in parts of the RAS research community.

Provide national testing facilities to demonstrate the viability of new methods and to promote take-up by industry.

Promote the development of regulation and accreditation of new RAS technologies to maximise economic and societal impact.

Ensure there is continuity of purpose between EPSRC, Innovate UK and relevant government departments for the development of RAS within key industry sectors.

Maintain an up-to-date profile of international excellence and specialisation in RAS across UK institutions.

Seek ways to drive innovation in academia through bringing new datasets into the public domain from major information-centric companies.

123

4

5

6

7

89

1011

1213

Ensure there is a mechanism in place to evaluate the early career pathways of alumni from the CDTs.

Review the success factors driving the participation of industry in the CDTs and distil practice for the future.

Review the balance between software and hardware research across all CDT PhD topics to inform the future direction of any potential RAS CDT programme.

CDTs that are part of the RAS portfolio should develop and share best practice on promoting gender diversity and collaborate in outreach activities to maximise impact. EPSRC should put in place a survey mechanism.

CDTs should consider how they could work together and with EPSRC to (i) signpost the RAS CDT portfolio to potential applicants, (ii) develop a national community of RAS PhD students, and (iii) harmonise and propagate best practice to promote mobility between CDTs.

141516

17

18

Robotics and Autonomous Systems Theme Day Panel Report 2017: Response from EPSRC

CHANGING LANDSCAPE

Since the Theme Day in January 2017, EPSRC has published refreshed research area rationales as part of its Balancing Capability strategy.1 The Robotics theme strategy3 outlined the need for this international benchmarking exercise and our goals for the area within this delivery period.The Theme Day approach was commissioned by the EPSRC’s Engineering Theme at a time of great change for public-funded research and innovation. The government’s Industrial Strategy and associated Industrial Strategy Challenge Fund (ISCF) altered the EPSRC-funded RAS landscape significantly. In a first wave of funding, government announced it would invest £93 million over four years to develop new technologies and systems that can be deployed in extreme environments.3

Implementing the recommendations of a review of the Research Councils by Sir Paul Nurse,4 the government have formed a new organisation: UK Research and Innovation (UKRI).5 From 1st April 2018, this will bring together the seven Research Councils, Research England and Innovate UK. The current transition period is already presenting an excellent opportunity to address these recommendations with one ‘funder’ voice.

EPSRC RESPONSE TO THEME DAY PANEL REPORT

EPSRC offers its thanks and appreciation to the panel members and participants whose assistance allowed for this evaluation to take place.

The Theme Day Panel’s recommendations (R), along with numbered paragraphs of context, were grouped by the themes of Quality, Impact, National Importance and Training. In recognition of the synergies

between the sections and to highlight specific issues, this response groups recommendations around the following themes: Quality and Impact, making links, Cross-TRL funding to maximise impact, Shared Infrastructure, International links and Training.

RESEARCH QUALITY AND IMPACT

RECOMMENDATION 1: EPSRC should work with the RAS community to identify critical gaps in investment across the whole UK RAS research landscape.

1. EPSRC is keen to support the RAS community by ensuring mechanisms are in place to support the input to such a gap analysis and how such information is fed into funders.

2. Through our Balancing Capability strategy, set out in our Strategic Plan6, we are committed to monitoring and reviewing the portfolio against the goals we laid out in our rationale for the Robotics area. We encourage the RAS community to engage with EPSRC on this through on-going discussions with our staff and the planned open call for new evidence to support our strategy review.

3. EPSRC has recently created a new role—Head of Robotics—to provide the leadership and coordination of the RAS portfolio. This will allow the organisation to ensure all the elements of the EPSRC portfolio—training, research, and innovation—are considered in future RAS funding decisions. Externally, the role also enables the organisation to engage with the wider research and innovation landscape and ensure EPSRC’s community is considered within emerging opportunities.

1https://www.epsrc.ac.uk/newsevents/news/balancingcapabilityresults/

2 https://www.epsrc.ac.uk/research/ourportfolio/researchareas/robotics/

3 https://www.gov.uk/government/collections/industrial-strategy-challenge-fund-joint-research-and-innovation#robotics-and-artificial-

intelligence-in-extreme-environments

4 https://www.gov.uk/government/collections/nurse-review-of-research-councils

5 https://www.ukri.org/

4. The UK currently contributes to the Multi-Annual Roadmaps funded by the EU Commissions’ Public Private Partnership between euRobotics Association Internationale Sans But Lucratif (AISBL) and their research network SPARC.7 EPSRC will work with the UK community through groups such as the EPSRC-funded UK-RAS Network and the Knowledge Transfer Network Robotics and Artificial Intelligence-Special Interest Group (KTN RAI-SIG) to ensure the UK research and business communities are involved in future activities.

5. EPSRC encourages the community to continue to drive evidence gathering to identify gaps in the RAS research landscape and draw on such evidence to generate and prioritise research projects and proposals. An example of such community-driven work is the series of white papers8 produced by the UK-RAS9 Network.

6. EPSRC plans to launch an open-to-all mechanism to help coordinate and validate ‘Big Ideas’ from universities and their user partners. The EPSRC Big Ideas activity will create a supportive environment for new ideas, discoveries and innovations arising from the research community to develop and flourish. Initially, EPSRC would welcome ideas particularly within the fundamental research and multidisciplinary areas to complement activities that are already in progress as part of the Industrial Strategy Challenge Fund.

RECOMMENDATION 10: Promote the development of regulation and accreditation of new RAS technologies to maximise economic and societal impact.

7. With the recent rapid growth of this area, EPSRC recognises the importance of striking the balance between ensuring regulation and accreditation is in place

to maximise impact and that there is the space for responsible innovation. As part of EPSRC’s Balancing Capability strategy for the RAS area, researchers are advised to take ethical considerations into account in all relevant situations.

8. EPSRC has recently provided funding to create a new organisation to deliver Responsible Research and Innovation services (RRI) called ORBIT.10 Hosted by De Montfort and Oxford Universities, they provide information and services to the ICT research and development communities to support them in undertaking their work responsibly. RRI is an approach that seeks to promote creativity and opportunities for science and innovation that are socially desirable and undertaken in the public interest.

9. In the Future Sectors section of the Industrial Strategy White Paper, the government outlines how new sectors and emerging businesses will be supported. Part of this will be through future Sector Deals, noting that the collaboration with government (academia and industry) will drive growth in new sectors such as artificial intelligence.

10. Where appropriate, EPSRC will continue to contribute to key RAS forums and coordinated national and international activities to ensure regulation and accreditation requirements are considered and prioritised. A number of reports have also recommended that a RAS Leadership Council111213 be formed in the UK and the intention would be for this group to play a key role in ensuring that aspects such as regulation, assurance and accreditation for RAS technologies are appropriately considered and addressed.14

11. Since the publication of the RAS Theme Day report, EPSRC have also been working closely with the Lloyd’s Register Foundation to understand the challenges around regulation in the RAS space. The Lloyd’s Register Foundation have also very

6 https://www.epsrc.ac.uk/about/plans/strategicplan/

7 A Brussels based international non-profit association for all stakeholders in European robotics https://www.eu-robotics.net/

eurobotics/topic-groups-/index.html

8 http://hamlyn.doc.ic.ac.uk/uk-ras/white-papers

9 http://www.uk-ras.org/

10 https://www.epsrc.ac.uk/research/ourportfolio/themes/ict/strategy/orbit/

recently funded the Assuring Autonomy International Programme which is being led by the University of York and is specifically focussed on addressing some of the questions regarding assurance of autonomous systems.

MAKING LINKS

RECOMMENDATION 2: Encourage research that links the strongest teams across the different fields that contribute to RAS.

RECOMMENDATION 3: Consider the development of inter-disciplinary and inter-technology connections, for example, promoting the interaction between robotics, machine learning and computer vision communities.

12. EPSRC encourages the research community to work proactively to draw on world-leading expertise to achieve their research goals. Mechanisms such as the EPSRC Programme Grant scheme encourage the community to bring the strongest teams together to tackle those challenges that need the interaction described in R3.

13. EPSRC will work with the UK-RAS Network15 to build on their now established leadership position to catalyse connections within and across other research communities. We are working with the leadership team to explore how best to support the community in the future and its fit within the current RAS landscape.

14. Through its Balancing Capability monitoring exercise (see response point 2) and consultation with the community, EPSRC will consider further potential interventions that could add significant value, in addition to taking ongoing advice from EPSRC Strategic Advisory Teams.Shared infrastructure

SHARED INFRASTRUCTURE

RECOMMENDATION 5: Consideration should be given to developing shared infrastructure for research in RAS, including the potential for industry to provide experimental facilities relevant to key sectors and challenges.

RECOMMENDATION 9: Provide national testing facilities to demonstrate the viability of new methods and to promote take-up by industry.

RECOMMENDATION 13: Seek ways to drive innovation in academia through bringing new datasets into the public domain from major information-centric companies.

15. The Research Councils take equipment sharing very seriously and EPSRC and JISC are currently supporting Equipment.data, a national equipment sharing portal, which publically-funded researchers are encouraged to contribute to.

16. The UK-RAS Network has set up a shared infrastructure database with over 200 separate items of equipment and/or facilities across 15 universities. The Network is using the list to help visualise the assets currently available to the academic and user communities and are currently looking into the benefits of a capital roadmap for RAS. The list includes the recent capital investment made in 2013 as part of the Eight Great Technologies investment.16

17. EPSRC, as a public funder of research, will continue to work with its strategic partners and wider user base to encourage the genuine co-creation of research ideas through mechanisms such as pre-competitive sharing of data. Examples of this best practice are visible

11 http://hamlyn.doc.ic.ac.uk/uk-ras/sites/default/files/RAS%20UK%20Strategy.pdf

12 https://www.nao.org.uk/wp-content/uploads/2017/11/Research-and-development-case-study-Robotics-and-autonomous-

systems-research.pdf

13 http://publications.parliament.uk/pa/cm201617/cmselect/cmsctech/896/89602.htm

14 https;//www.york.ac.uk/news-and-events/news/2017/research/lloyds-ras-partnership/

15 http://www.uk-ras.org/

throughout the EPSRC portfolio, such as the EPSRC Centre for Doctoral Training in Autonomous Intelligent Machines and Systems (AIMS) (EP/L015897/1).

18. A recent report by the National Audit Office17 and the RAS-SIG’s RAS 2020 Strategy both point to the need for national coordination and an investment in real world tangible RAS assets which are collaborative proving grounds. EPSRC supports this and will work with other key national bodies to ensure the EPSRC research community needs are considered in any future coordination activity.

INTERNATIONAL LINKS

RECOMMENDATION 6: The RAS research community and EPSRC should work to sustain and develop deeper international research links and joint funding opportunities, both within Europe and beyond.

RECOMMENDATION 12: Maintain an up to-date profile of international excellence and specialisation in RAS across UK institutions.

19. EPSRC’s Delivery Plan outlines how EPSRC are committed to developing international collaborations between the best researchers in the UK and those overseas.18 Recently this has been through targeted country interactions (currently Europe, USA, Japan, China and India) or through the Official Development Assistance (ODA) funding of the Global Challenge Research Fund.

20. Supporting this Delivery Plan ambition, EPSRC has piloted an open international call to support UK centres of excellence in bilateral partnerships with the best international collaborators. The call is aiming to support high-profile international collaborations focused on excellent research with impact. The call is open to all of EPSRC’s remit and to partnering with

all countries.

21. EPSRC will work with the RAS community with such opportunities and explore more specific opportunities as they emerge.

22. As part of EPSRC’s Balancing Capability

research area strategies, along with activities such as this Theme Day, we look at the international context of research areas.19 For robotics, a wide evidence base was used, including specific sources referencing the research landscape in both the USA and Japan.

23. The EPSRC ICT theme has also recently established an international sub-Strategic Advisory Team (ISSAT)20 to develop a theme-level strategy for engaging international groups. The group is considering how to best support leading UK-based ICT researchers to work with people and facilities across the world to create maximum mutual value.

CROSS-TECHNOLOGY READINESS LEVEL FUNDING TO MAXIMISE IMPACT

RECOMMENDATION 4: Consideration should be given to promoting collaboration between EPSRC-funded projects and to stimulating larger-scale projects at higher Technology Readiness Levels involving multiple institutions.

RECOMMENDATION 7: Ensure greater emphasis in the RAS portfolio on excellent foundational work, with a larger proportion of the wider portfolio linked to industrial and societal needs and opportunities, and positioned to translate into innovative application.

RECOMMENDATION 11: Ensure there is continuity of purpose between EPSRC, Innovate UK and relevant government departments for the development of RAS within key industry sectors.

16 https://www.epsrc.ac.uk/newsevents/news/85million/

17 https://www.nao.org.uk/wp-content/uploads/2017/11/Research-and-development-case-study-Robotics-

and-autonomous-systems-research.pdf

18 https://www.epsrc.ac.uk/about/plans/deliveryplan/internationalopportunities/

24. Since the Theme Day, the government has announced as part of the Industrial Strategy Challenge Fund (ISCF) the challenge area of Robots for a Safer World Challenge as one of the six challenges for ISCF’s first wave of funding.21 The vision of the ISCF challenge area is to have a connected support system for research, development, scale-up and industrialisation, with impact articulated at each stage. For the application-led research element, EPSRC used an open competition to commission four use-inspired Robotics and Artificial Intelligence (RAI) Hubs in Extreme and Challenging (Hazardous) Environments. This includes two centres looking at nuclear applications, one focused on space and another on offshore energy, whilst all are also encouraged to enable spill-over benefits to other sectors where appropriate.22 These highly collaborative centres of excellence involve 80 investigators and are working in direct partnership with industry with more than 100 project partners. The ISCF Robots for a Safer World Challenge is being delivered in partnership with Innovate UK who are delivering complementary investments in business led collaborative research and development and demonstrator projects, which is providing a coordinated approach to addressing the research and innovation aspects that are required to tackle this challenge.

25. The way in which the use-inspired RAI Hubs were commissioned required applicants to co-create their proposed plans with industrial partners and end users of the research to ensure that there was a genuine and appropriate pathway to impact for the research outputs, which is essential to ensure that the challenge is addressed and that both the industrial and societal impact is maximised.

26. EPSRC will work with Innovate UK to provide coordinated support for the ISCF-funded RAS portfolio (i.e. the use-inspired RAI Hubs, Innovate UK-

funded Collaborative R&D proposals and Demonstrator projects) to look for synergies, cross-sector learning, encourage sector spill-over and to bring our best RAI researchers together.

27. This coordinated approach will be further supported through the formation of UK Research and Innovation (UKRI) and the development of a UKRI position statement for this area during 2018.

RECOMMENDATION 8: Seek ways to replicate the successful innovation mechanisms that exist in parts of the RAS research community.

The report states (Para 24): “There are outstanding examples of ‘innovation engines’ driven by individual researchers from within the UK RAS community.” And “Consideration should be given to new mechanisms or existing schemes that can empower and motivate other leading researchers to adapt and replicate these successes. For example, allowing a period of time in which a researcher can focus on building a team, its competencies and infrastructure, without the distraction of having to devote time and effort to raising funds, enables a project to start on a strong footing from which innovation can thrive.”

28. EPSRC’s Delivery Plan states that we will maintain a programme of long-term, excellent research where the emphasis is on ‘bottom-up’ investigator-led ideas, including community-generated challenges. Research Excellence Framework (REF) Impact Case Studies show that real-world benefits often emerge over a twenty-year timeframe.

29. Such long-term capability building is achieved in the portfolio through mechanisms such as EPSRC’s Programme Grant scheme and EPSRC’s Fellowship scheme, in which RAS is a cross-cutting

19 See the Influences tab within https://www.epsrc.ac.uk/research/ourportfolio/researchareas/robotics/

20 First meeting attendees (20 September 2017): Weiru Liu, Tony Kenyon, Edmund Burke, Bashar Nuseibeh, David Robertson,

Yvonne Rogers, Yang Hao, Martin Kuball, Chris Johnson.

21 It aims to develop robotics and artificial intelligence technologies that can take people out of dangerous work environments and

go beyond human limits.

22 https://www.epsrc.ac.uk/newsevents/news/break-through-robotics-ai-funded-iscf/

23 33 universities currently hold IAAs, the latest list was announced with the Prosperity Partnerships (see previous link)

open theme for applicants to apply at any time. These mechanisms are specifically designed to give long-term support (typically 5 years) to our research leaders, thus allowing space for innovation and impact to flourish.

30. For these and standard mode proposals, EPSRC applicants are required to use the Pathways to Impact section of their proposal to identify the potential impact of their work and to outline the steps they can sensibly make now to facilitate the realisation of those impacts. This encourages researchers to think about impact at an early stage and ask for the resources to achieve it.

31. Our institutionally awarded Impact Acceleration Accounts (IAAs), which have recently been refreshed for 3 years, complement this project-based approach23. The IAAs enable institutions to react to unexpected opportunities to progress their EPSRC-funded research towards impact, by giving them the flexible support to bridge funding gaps and support a range of Knowledge Exchange (KE) activities that best suits their strategies and opportunities.

32. These allow institutions–either on their own or as part of a consortium–the flexibility to operate tailored schemes that help increase the likelihood of impact from their research.

33. EPSRC will continuously monitor and review its mechanisms, including the Theme Day Report’s reference (para 22) to the Towards Autonomy – Smart and Connected Control (TASCC) initiative.24 This initiative involves working to pre-competitive goals, set in partnership with a user that will benefit those beyond the partnership.

34. EPSRC has since expanded this partnership model with the Prosperity Partnership call. This co-funding scheme’s aim is to support existing, strategic, research-based partnerships between businesses and universities.

The pilot call run in 2016 has funded eleven such partnerships and a second call is currently underway.25

TRAINING

RECOMMENDATION 14 : Ensure there is a mechanism in place to evaluate the early career pathways of alumni from the CDTs.

35. The Theme Day covered four centres as part of the review: three CDTs that were awarded as part of the 2013 call and one previously awarded as an Engineering Doctorate programme. These centres started with their first intakes in October 2014 and thus graduates are only starting to emerge in 2018.

36. As part of the monitoring and evaluation process, EPSRC collects the first six-month destination data from all of the CDTs via the HESA reporting mechanism.

37. EPSRC is aware of the need to collect more longitudinal data on our funded students and is currently exploring ways of tracking the destinations of graduates on a longer timescale. As part of this we are asking universities for any further data sets they could provide on the destination of their postgraduate students. EPSRC and HESA are also exploring how EPSRC CDTs could be flagged in future longitudinal studies.

38. EPSRC recognises the need for more local, refined capture of follow-on student behaviour at CTDs, and this will be investigated further through mechanisms such as LinkedIn groups. Experience at the Research Councils suggests that there is better student take-up of such networks if they are self-organised rather than funder-driven.

RECOMMENDATION 15: Review the success factors driving the participation of industry in the CDTs and distil practice for the future.

24 Five proposals funded through a strategic partnership call with Jaguar Land Rover: http://gow.epsrc.ac.uk/NGBOViewPanelROL.

aspx?PanelId=1-2UVNCT&RankingListId=1-2UVND1

25 Eleven partnership awards: https://www.epsrc.ac.uk/newsevents/news/partnershipsprosperousnation/

26 https://www.epsrc.ac.uk/newsevents/news/cdtreview/

39. Following a mid-term review of current EPSRC portfolio of centres in spring 2017,26 EPSRC recently published a set of case studies of CDT engagement “Building skills for a prosperous nation” (https://www.epsrc.ac.uk/newsevents/pubs/cdtprosperousnation/) that has 18 mini-case studies of how users—public and private sectors—have benefited from the cohort training approach.

40. EPSRC is currently collating a series of different models for CDTs that illustrate the different methods of delivery that have worked, which will be disseminated via our website.

RECOMMENDATION 16: Review the balance between software and hardware research across all CDT PhD topics to inform the future direction of any potential RAS CDT programme.

41. EPSRC supports the research and user community identifying the optimum balance required between software and hardware research to address the skills gap they identify. The 2018 CDT call will be open to the community to shape the training to provide the skills required to fulfil the UK’s potential in this area. The RAS-SIG’s strategy document, RAS 2020 Robotics and Autonomous Systems (July 2014), highlighted RAS skills as a key area requiring coordination, and the EPSRC-funded UK-RAS Network is one way this is being done.

RECOMMENDATION 17: CDTs that are part of the RAS portfolio should develop and share best practice on promoting gender diversity and collaborate in outreach activities to maximise impact. EPSRC should put in place a survey mechanism.

42. RCUK’s Action Plan for Equality, Diversity and Inclusion (E,D&I) forms the framework for EPSRC’s work in promoting best practice in providing an inclusive research and training environment as part of business as usual.

43. EPSRC asks all existing CDTs as part of our evaluation and monitoring process to provide evidence that equality, diversity and inclusion has been in built to their centres.

44. There was a strong recommendation from the mid-term review for all of the centres to show proactivity in promoting E,D&I. We want to encourage centres to understand the issues beyond collecting data and to show evidence that positive action has been taken. Our vision is for the CDTs to be leading examples of best practice across both the training and research environments. The CDT 2018 call will look for this explicitly.

RECOMMENDATION18: CDTs should consider how they could work together and with EPSRC to (i) signpost the RAS CDT portfolio to potential applicants, (ii) develop a national community of RAS PhD students, and (iii) harmonise and propagate best practice to promote mobility between CDTs.

45. EPSRC encourages this type of coordination across of all its CDTs and believes the RAS community is already making good head way with this through the UK-RAS Network with events such as the student conference. EPSRC will continue to act as a broker between centres where appropriate and help in sharing best practice.

27 http://www.rcuk.ac.uk/funding/diversity/

Annex 1: This Response’s grouping ofrecommendations

R Recommendations by original grouping Response groupings

Response theme

Research capability

1 EPSRC should work with the RAS community to identify critical gaps in investment across the whole UK RAS research landscape.

1,10 Research Quality and Impact

2 Encourage research that links the strongest teams across the different fields that contribute to RAS.

2,3 Making Links

3 Consider the development of inter-disciplinary and inter-technology connections, for example, promoting the interaction between robotics, machine learning and computer vision communities.

2,3 Making Links

4 Consideration should be given to promoting collaboration between EPSRC-funded projects, and to stimulating larger-scale projects at higher Technology Readiness Levels involving multiple institutions.

4,7,11 Cross-Technology Readiness Level funding to maximise impact

5 Consideration should be given to developing shared infrastructure for research in RAS, including the potential for industry to provide experimental facilities relevant to key sectors and challenges.

5,9,13 Shared infrastructure

6 The RAS research community and EPSRC should work to sustain and develop deeper international research links and joint funding opportunities, both within Europe and beyond.

6,12 International links

Impact (Academic and Economic and Societal)

7 Ensure greater emphasis in the RAS portfolio on excellent foundational work, with a larger proportion of the wider portfolio linked to industrial and societal needs and opportunities, and positioned to translate into innovative application.

4,7,11,8 Cross-Technology Readiness Level funding to maximise impact

8 Seek ways to replicate the successful innovation mechanisms that exist in parts of the RAS research community.

4,7,11,8 Cross-Technology Readiness Level funding to maximise impact

9 Provide national testing facilities to demonstrate the viability of new methods and to promote take-up by industry.

5,9,13 Shared infrastructure

10 Promote the development of regulation and accreditation of new RAS technologies to maximise economic and societal impact.

8,10 Research Quality and Impact

11 Ensure there is continuity of purpose between EPSRC, Innovate UK and relevant government departments for the development of RAS within key industry sectors.

4,7,11 Cross-Technology Readiness Level funding to maximise impact

12 Maintain an up-to-date profile of international excellence and specialisation in RAS across UK institutions.

6,12 International links

13 Seek ways to drive innovation in academia through bringing new datasets into the public domain from major information-centric companies.

5,9,13 Shared infrastructure

Training

14 Ensure there is a mechanism in place to evaluate the early career pathways of alumni from the CDTs.

14,15,16,17,18 Training

15 Review the success factors driving the participation of industry in the CDTs and distil practice for the future.

14,15,16,17,18 Training

16 Review the balance between software and hardware research across all CDT PhD topics to inform the future direction of any potential RAS CDT programme.

14,15,16,17,18 Training

17 CDTs that are part of the RAS portfolio should develop and share best practice on promoting gender diversity and collaborate in outreach activities to maximise impact. EPSRC should put in place a survey mechanism.

14,15,16,17,18 Trainng

18 CDTs should consider how they could work together and with EPSRC to (i) signpost the RAS CDT portfolio to potential applicants, (ii) develop a national community of RAS PhD students, and (iii) harmonise and propagate best practice to promote mobility between CDTs.

14,15,16,17,18 Training

Annex 2: Other reviews with relevance to RAS

Report title Author(s) Notes

RAS 2020 Robotics and Autonomous Systems (July 2014)

Innovate UK A national strategy to capture value in a cross-sector UK RAS innovation pipeline through co-ordinated development of assets, challenges, clusters and skills.

UK-RAS Network White Papers series, with the latest 4 (*) published in June 2017:

•Artificial Intelligence and Robotics*•Robotic and Autonomous Systems for Resilient Infrastructure*•Extreme Environments Robotics: Robotics for Emergency Response, Disaster Relief and Resilience*•Robotics in Social Care: A Connected Care EcoSystem for Independent Living*•Surgical Robotics: The Next 25 Years Successes, Challenges, and the Road Ahead•Manufacturing Robotics•The Next Robotic Industrial Revolution

UK-RAS Network The UK-RAS Network has published a series of white papers covering key areas within robotics and autonomous systems.

The UK Landscape for Robotics and Autonomous Systems (PDF), (2015).

Robotics and Autonomous Systems Special Interest Group (RAS SIG)

This report captures the outcomes of the RAS SIG’s hosted series of workshops designed to explore the impact on the UK of advances in RAS.

Made Smarter. Review 2017 Industrial Digitalisation Review, http://industrialdigitalisation.org.uk/

This report summarises the findings and recommendations of the Made Smarter Review (previously referred to as the Industrial Digitalisation review), which was announced in the Industrial Strategy Green Paper in January 2017.

Industrial Strategy and RAS Sector deal consultation:https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/662508/industrial-strategy-white-paper.pdfandhttps://ktn-uk.co.uk/news/robotics-sector-deal-consultationand

The Robotics Sector Deal Proposal Team

The Robotics Sector Deal Proposal Team In the Future Sectors section of the government’s Industrial Strategy White Paper, the government outline how new sectors and emerging businesses will be supported. Part of this will be through future Sector Deals with emerging and disruptive industries where there is a compelling case to do so, beginning with artificial intelligence. The collaboration with government will drive growth in new sectors such as robotics and artificial intelligence

Cross-government funding of research and development and the case study within:Research-and-development-case-study-Robotics-and-autonomous-systems-research

(The report also examined five other research areas: human health; animal and plant health; climate; energy; and advanced materials.)

National Audit Office, November 2017

With the recent changes to public funding of R&D in the UK, the NAO have published a report on how this is coordinated, prioritised and reviewed. They used an evaluative framework to assess the effectiveness of arrangements for coordinating research activity and maximising the value of government’s investment in research. They compared arrangements across six research areas, of which RAS was one, in order to highlight good practice and identify where improvements could be made.

Robotics and artificial intelligence: Government Response to the Committee’s Fifth Report of Session 2016–17

Dec 2016 Includes a recommendation to establish a RAS Leadership Council.

Engineering and Physical Sciences Research CouncilPolaris HouseNorth Star AvenueSwindonSN2 1ET

www.epsrc.ac.uk