Issue link: https://htpgraphics.uberflip.com/i/1385717
LEAD RESEARCHER: THOMAS WRIGHT I am a Post Doctoral Research Associate working at the University of Manchester as part of the RAIN Hub. The main focus of my research has been to develop and integrate technologies and algorithms which allow for robots to map and characterise nuclear environments. The RAIN Hub has supported this research to allow for continued investigation of this subject, whilst also making possible collaborations with other researchers interested in the same field from the wider robotics community. Being part of the RAIN Hub has helped publicise these tools to the wider nuclear robotics industry, leading to many researchers both inside RAIN hub and external to adopt these tools. This has helped shape the tools to make them more broadly applicable and able to support more researchers work. SIMULATION TOOLS FOR EXPLORATION AND CHARACTERISATION OF NUCLEAR ENVIRONMENTS SUMMARY // This work comprises two main parts which share a common motivation. 1. World builder toolbox: It is currently a laborious and tedious task to create a simulated 3D environment for a robot to explore. This can severely limit the amount of testing which is done in simulation. Due to the nature of nuclear facilities, it is highly desired that a robot be exhaustively tested as the occurrence of undesired or unexpected behaviours could have heavy ramifications. This tool aims to simplify the process of generating a number of unique scenarios to offer a diverse set of test environments. 2. Simulated radiation sensors and sources. It is even more complex to include radiation characteristics in simulations and have algorithms/ robots react to its presence. Gazebo – a commonly used simulation platform- has no simple option to quickly implement features outside of its existing feature set suite. This meant adding radiation sensors and sources was a desirable piece of work as without it, it would be difficult to robustly test in simulation, algorithms that need to respond to the presence of radiation. Current tools/approaches, such as GEANT4 and MCNP provide high levels of representability but are not designed for use with mobile robots or the control algorithms they run. They also require specialist knowledge and are not designed for near real-time dynamic simulations. 38