HTP Graphics

LEAD RESEARCHER: BRUNO LACERDA I am a Senior Researcher at the Oxford Robotics Institute (ORI). I specialise in formal methods for robotics and AI, with a focus on decision-making under uncertainty. In particular, I focus on the use of Markov decision processes, and extensions thereof, to model robots and their environments; and rich specification languages, such as temporal logics, to specify rich behaviour over these Markov models. This allows for providing performance guarantees over the synthesised behaviour. These guarantees can formally represent key notions of robot behaviour, such as safety, robustness or efficiency. Working in RAIN gives me a rich and challenging domain for applying my work, given the extreme and safety-critical nature of nuclear environments. SUMMARY // My work in RAIN has dealt with mission planning for inspection tasks in nuclear environments. I am developing approaches that consider the trade-off between inspecting as much of the environment as possible and the amount of radiation the robot is exposed to. For my work this year, I focused on situations where the location of radiation sources in the environment is unknown a-priori. The goal is to efficiently explore the environment, whilst at the same time ensuring safety, i.e. that the robot does not navigate to areas with high radiation. In order to do so, the robot must use its sensor readings to maintain probabilistic models of radiation across the workspace. These models should generalise to areas that have not been visited yet, and include a notion of how confident the robot is about the radiation in those areas. This allows the decision- making mechanism to intelligently consider where to explore next, such that safety is maintained whilst the environment is efficiently explored. MISSION PLANNING UNDER UNCERTAINTY IN NUCLEAR ENVIRONMENTS 28

• Cover