At PR #20 head 1c2686b, recon/recon/coverage.py:163-175 samples ground on the horizontal plane through the meter's ground height. geometry.py:260-264 evaluates the fitted ground only at the meter, so the outward samples do not follow the fitted slope.
On ground sloping away from the wall, once the actual surface differs from that horizontal sample by more than the depth tolerance, visible ground becomes unobserved and the server returns UNSURE. No false-clear outcome was demonstrated from this mismatch alone.
Pass the fitted ground plane into coverage and evaluate each sample's local height. Add an asymmetric sloped-ground case showing that observed surface samples follow the plane while genuinely unseen samples remain unobserved.
Source: #20 (comment)
Deferred by the manager to Hunter's team as a P2 follow-up for the draft reconstruction handoff.
At PR #20 head
1c2686b,recon/recon/coverage.py:163-175samples ground on the horizontal plane through the meter's ground height.geometry.py:260-264evaluates the fitted ground only at the meter, so the outward samples do not follow the fitted slope.On ground sloping away from the wall, once the actual surface differs from that horizontal sample by more than the depth tolerance, visible ground becomes unobserved and the server returns UNSURE. No false-clear outcome was demonstrated from this mismatch alone.
Pass the fitted ground plane into coverage and evaluate each sample's local height. Add an asymmetric sloped-ground case showing that observed surface samples follow the plane while genuinely unseen samples remain unobserved.
Source: #20 (comment)
Deferred by the manager to Hunter's team as a P2 follow-up for the draft reconstruction handoff.