Abstract
Safe, cost-effective inspection of nuclear facilities is hampered by strict human dose limits, awkward geometries and the tendency of multirotor drones to resuspend radioactive dust. This thesis examines whether lightweight, off-the-shelf sensors mounted on buoyant non-rigid airships can deliver reliable indoor localisation, mapping and radiometric characterisation while tolerating ionising radiation.A comprehensive literature review compares acoustic, optical and radiation-detection techniques and motivates a sensor suite composed of a planar LiDAR, an Intel RealSense T265 tracking camera and a cadmium-zinc-telluride (CZT) gamma spectrometer, fused in real time with the Cartographer ROS SLAM framework.
Experimental work proceeded in four strands. First, twenty VL53L1X time-of-flight modules were exposed to 95 Gy h⁻¹ of γ radiation for 48 h; none exhibited measurable drift, demonstrating short-term hardness far beyond expectations for consumer optoelectronics. Secondly, down-sampling Velodyne VLP-16 data showed that scan matching remained stable with about fifteen independent beams per revolution. A decagonal ring of ten VL53L1X sensors was built and tested, providing a near-360° field of view at 20 Hz over Bluetooth Low Energy for collision avoidance. Thirdly, a two-metre helium blimp carrying a vertically mounted RPLIDAR S1 and the T265 generated dense three-dimensional maps of an indoor warehouse; Cartographer’s sub-map and pose-graph optimisation achieved accurate three-dimensional maps without the mass of a spinning 3-D LiDAR. Finally, the blimp payload was augmented with a 60 g Kromek GR1 CZT detector. Gamma counts linked to pose were inverted, using a Kaczmarz solver and detector-response model, to colour the point cloud with dose rate and successfully localise a 370 kBq ¹³⁷Cs source.
These results demonstrate that inexpensive, lightweight sensors can be orchestrated into radiation-tolerant, multimodal systems for inspection. The thesis contributes (i) a systematic γ-hardness characterisation of VL53L1X modules, (ii) a pragmatic beam-density guideline for sparse LiDAR arrays, (iii) a lightweight SLAM pipeline for non-rigid airships and (iv) an integrated geometric–radiometric mapping framework, thereby opening practical pathways to safer and more economical robotic operations in radiation-enriched environments.
| Date of Award | 9 Dec 2025 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Thomas Bligh Scott (Supervisor), John C C Day (Supervisor) & David Megson-Smith (Supervisor) |