Optimal Measurement Strategies for 3D Dose‑Field Reconstruction in Hazardous Environments

Background

Radiation‑protection workflows such as hot‑lab inspections, decommissioning, emergency response, and accelerator operations often take place in complex environments with shielding, activated components, and high‑gradient dose fields. Computing full‑resolution 3D radiation fields can be slow or impossible, and VRdose frequently depends on interpolated fields from sparse measurements.

However, the quality of reconstruction strongly depends on where measurements are taken — and this is further constrained by who or what can perform measurements in hazardous areas (humans, robots, drones).

Goal

Develop a prototype that determines optimal measurement locations for reconstructing accurate 3D dose fields under agent constraints and compare the results directly with VRdose’s built‑in interpolation.

Key Research Questions

1. How do different sampling strategies affect reconstruction accuracy in complex radiation fields?
2. How do measurement‑agent constraints (human, robot, drone) impact feasible measurement locations?
3. How does an optimized reconstruction compare to VRdose’s interpolation when using identical sparse datasets?

Objectives

• Model synthetic “ground‑truth” dose fields using simple sources and shielding.
• Implement agent‑aware sampling strategies:
  • Human: limited to safe paths, low‑dose corridors, reachable heights.
  • Robot: access to tighter and higher‑dose areas, high‑gradient zones.
  • Drone: 3D volumetric sampling with obstacle avoidance.
• Reconstruct the dose field from sparse samples using methods like RBF, kriging, Gaussian processes, or adaptive meshes.
• Import the same measurement points into VRdose, let it interpolate, and compare both reconstructions to the ground truth.
• Identify sampling patterns that provide high accuracy for VRdose‑relevant workflows.

Methodology (Workflow)

1. Generate synthetic dose fields (Python/NumPy).
2. Apply sampling strategy based on agent constraints.
3. Reconstruct the 3D field externally.
4. Import sample points into VRdose → obtain VRdose interpolated field.
5. Compare accuracy: error maps, gradient errors, path‑integrated dose.
6. Produce measurement‑planning guidelines for realistic scenarios.

Expected Outcomes

• A prototype “measurement planner” for hazard‑aware sampling.
• Comparative accuracy study vs. VRdose interpolation.
• Guidelines for optimal measurement strategies in nuclear workflows (inspection, decommissioning, emergency operations).
• Results directly usable in future VRdose pipeline improvements.

Why This Is Valuable

• Scientifically rigorous, publishable results.
• Direct relevance to VRdose workflows involving complex geometries and dose‑field optimization.