Written Evidence Submitted by

Professor Andrew Davison, FREng, Department of Computing, Imperial College London

(RFA0006)

Views on a New UK Funding Agency for Science and Technology

I am one of the UK’s most highly cited mid-career academics in AI, focusing on robotics and computer vision. I have also had significant industry impact, including working with Dyson to design the vision system inside the Dyson 360 Eye robot vacuum cleaner which was based on my academic MonoSLAM work. I currently direct the Dyson Robotics Lab at Imperial College. I am also the co-founder of London startup SLAMcore, and my PhD students co-founded Surreal Vision, acquired by Facebook in 2015.

My views on scientific funding are clear. The UK has a strength in deep science and true innovation, but this risks becoming increasingly weakened by the increasing focus on funding which reinforces existing large research groups (even though I myself have recently benefitted from things like this such as the EPSRC Prosperity Partnership scheme), and government-decided research directions, which lead to politically minded researchers building large groups, centres and consortia which are very good at getting the lion’s share of funding and playing a very strategic research game which has only very incremental scientific benefits.

I believe that the best science, leading to the small number of true advances which ultimately have real economic or societal benefit, comes from curiosity-driven, long-term research. The best ways of funding this in my opinion are to give it as many chances as possible to happen in universities throughout the land. The most important mechanisms are:

•    Individual fellowships, both for early career (post-doctoral) and later career researchers, along the lines of EPSRC or RAEng 5 year fellowships which still exist but are getting harder to compete for. These give individuals the chance to have real research freedom, and can be key to attracting and retaining top talent in UK universities. In my early career I had a 5 year EPSRC Advanced Research Fellowship as I transitioned from post-doc to lecturer and I did the best research of my career (which directly led to the work with Dyson) in this period. Later on I had an ERC Starting Grant which played a similar role for 5 years when I was just founding my research group.

•    PhD studentships, of a plentiful, non-directed, and widely available nature. I don’t believe in schemes like Doctoral Training Centres, which are positive feedback schemes and focus funding on a few big centres. Make it easy for any talented young lecturer to have access to PhD studentships to be able to offer to the talented people who come along and want to work with them, with no strings attached on the work they are required to do. Ideally these should be available to offer to students of any nationality, because it is currently very hard for us to hire overseas PhD students and there is a huge talent pool available that want to come and work in the UK. PhD students are the core of any thriving research group.

•    Responsive mode grants. We need just good old simple responsive mode such as is still available from EPSRC for any type of grant, but the pot has got smaller and smaller in recent years as more funding has gone into focused schemes with positive feedback towards well known strategic groups. Put the money back into responsive mode and every lecturer with a good idea will have more chance to put in a high probability bid. The pot is so small at the moment that a huge amount of effort is being wasted by researchers writing good grant proposals with very low probability of being funded.

I strongly believe that strengthening these simple measures will have a much bigger positive effect on retaining deep science and innovation in the UK than any kind of strategic schemes or moonshot projects. Make UK universities the most attractive place for talented young scientists to work in an unfettered, low overhead way on their own self-driven research.

(June 2020)