Skip to main content

EPSRC announces grants for robotics research projects

Posted:

Research that could enable groups of flying or underwater robotic vehicles to work together more effectively has received a £16m funding boost. The EPSRC today announced grants for 22 university-based robotics research projects supported by a number of major British companies and organisations including BAE Systems, Network Rail and the UK Space Agency.

Daniel Kroening of Oxfords’ Department of Computer Science is leading research into co-ordinating groups of UAVs, which he hopes could improve the way they work when collecting data in difficult conditions, for example during a Fukushima-style disaster.

The researchers want to develop algorithms that would help UAVs to share information and make group decisions and to enable them to anticipate each other’s movements if communication fails. By bringing together mathematicians and engineers, the project also aims to develop ways to test the programming in both computer simulations and in real models, in order to prove to aviation authorities that it is safe.

Daniel stated ‘The working-together aspect is very important once you have groups of UAVs and it’s something that puts more stress on the reasoning and decision-making capabilities because you need to anticipate decisions made by other UAVs,’  

The algorithms will address situations where communication and group decision making is needed, for example when access to GPS is limited or where one UAV in the group stops working and the rest need to replan the mission.

‘This would be impossible to do if you were to give each of these UAVs individual programming — they would not talk to each other,’ he continued.

But the software will also enable UAVs to guess what their colleagues are doing when communication fails — something that requires a very complex level of programming. This idea of computers thinking about other computers’ thinking will also come into play in the validation models the researchers want to create, which are needed because the software is likely to be too complex to easily prove its safety.

‘We use a computer to consider all possible outcomes of these thinking processes ,,,, So the computer tries to find flaws in the thinking of the UAV, and by not finding any we can conclude that the high-level system of the UAV is safe to operate.’

The £16m fund represents new investment by the EPSRC that, together with £4m contributed by the industrial partners, was allocated because of the opportunity for commercially exploitable technology presented by the research. Among the projects is a notable focus on technology related to monitoring large, hard-to-reach structures such as nuclear power stations using autonomous networks of unmanned aerial vehicles (UAVs) or low-cost sensors.