Pose Estimation for Mobile Robots to Maximise Data Quality of Fixed-Focus Laser Diagnostics in Hazardous Environments

Abstract

Characterisation of nuclear environments is critical for long term operation and decommissioning. Laser Induced Breakdown Spectroscopy (LIBS) is an example of a scientific instrument that could be deployed to aid in characterisation of unknown environments. LIBS consists of a high intensity pulsed laser being focussed down onto a target to create a plasma, and optical emission from the plasma is then used to determine elemental composition of unknown materials. For robots deployed with these instruments in extreme environments, mission time can be limited by hazards present such as radiation. Once deployed a robot must be able to collect the best data possible whilst maximising operational runtime. We present a data quality based probabilistic approach to robot pose estimation to maximise data quality, by considering optimum sensor placement whilst avoiding harmful environmental features such as radiation for a fixed-focus laser diagnostic such as LIBS. This approach is able to determine optimum robot poses for arbitrary targets in 3D for arbitrary diagnostic mounting with respect to the robot. The approach is able to avoid obstacles and avoid occlusion of the target by said obstacles. This can be used as part of autonomous investigation and characterisation performed by mobile robots in hazardous environments.