Large volume metrology is a key component of autonomous precision manufacturing. Photogrammetry systems are an example of an optical-based metrology system and can be used for component positioning.However,these positional measurements are subject to uncertainties which can be greater than manufacturing tolerances. In large scale industrial environments the uncertainties due to thermal gradients which cause refraction of the light rays, need to be considered. This paper uses light-based sensor data to reconstruct the heterogeneous spatial map of the refractive index in the air. This is then used to discount the refractive eﬀects and thereby reduce the uncertainty of this positioning problem. This new inverse problem employs Voronoi tessellations to spatially parameterise the refractive index map, the forward problem of calculating the light rays through this medium is solved using the Fast Marching Method, and a Bayesian approach is then used as the optimisation method in the inversion. Using simulated data, the recovered refractive index map leads to positioning improvements of up to 37 %.
|Journal||Inverse Problems in Science and Engineering|
|Publication status||Submitted - 2020|
- refractive index
- reversible jump Markov Chain Monte Carlo
- Fast Marching Method
- Voronoi tessellations