Q1. What model scale (may need a new model) and model "roughness" would be required to see a difference in sonar data?
Q2. Create a demonstration of A undegraded and B degraded (3D degradation) and show that the sonar data looks "different" - passing the sniff test.
Classification for the definition of ‘difference’ in the sonar image
Macroscopic geometry
Changes in object size
Acoustic parameters
Changes in the acoustic parameters in the sonar (i.e. reflectivity)
Deviation in incident angles (which affects amplitude and phase)
Microscopic material properties
surface roughness
Note that currently, this is out of the scope for object degradation. There is a rabbit hole we can explore to implement for the seabed. The vocab. ‘roughness’ used in the seabed scattering is something that distinguishes silt and sand.
Classification for the definition of ‘difference’ in the sonar image
Discussions
Live Document link
Demonstration of rough features
https://github.com/Field-Robotics-Lab/nps_uw_multibeam_sonar/pull/33#issuecomment-1046747184
Multibeam Sonar Wiki - Degraded object detection scenarios
Distortion mesh statistics discussions
https://github.com/Field-Robotics-Lab/dave/issues/187