feat: qualify K1 local surface over time

docs/13_LIDAR_WORKER_PRODUCT_AND_ROADMAP.md

@@ -2,8 +2,9 @@
 
 Date: 2026-07-25
 Status: accepted architecture plan; L0/L1 implemented; L2 diagnostic A/B
-complete; full GOOSE and RELLIS qualification complete; L2.6a K1 replay
-local-surface slice implemented; operator review and live shadow next
+complete; full GOOSE and RELLIS qualification complete; L2.6b K1 replay
+local-surface temporal qualification implemented; operator review and live
+shadow next
 Scope: passively received real-time K1 point/pose evidence, immutable replay and
 future live shadow processing
 Explicitly out of scope: K1 firmware modification, a new onboard exporter, new
@@ -398,7 +399,7 @@ Dataset expansion is no longer the next gate.
   `1.27 m` value.
 - [x] Publish surface-relative height, slope, roughness and confidence
   separately from semantic classes.
-- [ ] Add independently reviewable step/curb candidates; do not derive them
+- [x] Add independently reviewable step/curb candidates; do not derive them
   from a single global height threshold.
 - [x] Produce short-TTL observed-surface, `occupied` and `unknown` evidence.
   Do not infer `free` merely because a mapped point is absent.
@@ -408,26 +409,47 @@ Dataset expansion is no longer the next gate.
   derivatives with independent decay and provenance.
 - [ ] Reuse the accepted camera-to-LiDAR projection as an optional semantic
   layer with source, confidence, freshness and conflict fields.
-- [ ] Complete the qualification report with per-frame latency, point age,
-  temporal stability, obstacle preservation and memory growth.
+- [x] Qualify next-frame prediction and temporal stability with the evaluated
+  frame excluded from prediction input.
+- [ ] Complete the remaining qualification report with per-frame latency,
+  point age, obstacle preservation and memory growth.
 - [ ] Replay the same profiles through a bounded latest-wins live-shadow queue;
   no K1 command, navigation or safety authority is added.
 
 The implemented `missioncore.k1-local-surface/v1` derivative is reproducible
 through `experiments/perception/run_k1_local_surface.py` and is exposed
 read-only through `GET /api/v1/lidar/local-surfaces` plus the bound frame
-endpoint. **Парк → Диагностика LiDAR** reuses the five RAVNOVES00 scene
-selectors and shows the selected source frame as observed surface, observed
-occupied-above-surface, negative outlier and unclassified evidence. The React
-contract is provider-neutral; K1 remains a bound backend source rather than UI
-implementation knowledge.
+and timeline endpoints. **Парк → Диагностика LiDAR** reuses the five
+RAVNOVES00 scene selectors and also exposes a clickable timeline over the
+complete recording. The selected source frame shows observed surface, observed
+occupied-above-surface, negative outlier, unclassified evidence and yellow
+local-discontinuity candidates. The candidates are not semantic curb
+recognition or ground truth. The React contract is provider-neutral; K1 remains
+a bound backend source rather than UI implementation knowledge.
 
 The complete RAVNOVES00 run covered all `526/526` available LiDAR samples.
 There were zero stale-pose, insufficient-surface or failed-fit frames. Observed
 diagnostic distributions are: derived sensor-to-surface height `1.295 m` p50,
 roughness `0.054 m` p95, slope `3.172°` p95 and pose-binding age `19.113 ms`
 p95. These values describe this recording only; they are not calibration,
-ground truth or a navigation gate. Free space remains unavailable.
+ground truth or a navigation gate.
+
+The L2.6b qualification scores each available frame against a local plane built
+only from the preceding TTL window; the frame being scored is excluded from
+the prediction input. It produced `525` independent next-frame samples. The
+distribution of per-frame median absolute residual has `0.038543 m` p50 and
+`0.050398 m` p95. Prediction inlier fraction has `0.96354` p50 and `0.95133`
+mean. The temporal comparison detected `12/525` strict-threshold jumps; height
+delta is `0.022887 m` p95, slope delta `0.09572°` p95 and roughness delta
+`0.003317 m` p95.
+
+The central surface is therefore repeatable at roughly four-centimetre median
+error on this replay. The tail is not yet planner evidence: the distribution
+of per-frame p95 residual has `0.41496 m` p95 and reaches `1.083 m`. Step/curb
+candidates occur in every available frame (`68.5` cells and `83` points p50),
+which proves that the review layer is active but also that it is broad and
+still requires independent review. Free space remains unavailable; none of
+these metrics grants navigation, command or safety authority.
 
 Exit: one immutable K1 session yields both a persistent reconstruction and a
 bounded local world state without hard-coded terrain height or scanner-side
@@ -512,8 +534,10 @@ and rejects Patchwork++ because the small Ground-IoU gain came with unacceptable
 obstacle loss. Public-dataset ground qualification is therefore complete
 enough for the current decision. The first K1 local-surface replay slice now
 covers all available `RAVNOVES00` samples and is visible in the operator
-interface. The highest-value immediate work is operator review, explicit
-step/curb and temporal-stability qualification, followed by bounded live
-shadow. Nvblox, raw-scan detectors and alternative SLAM remain optional later
-gates because the current report contract does not carry their required
-ray/timing semantics.
+interface. Leave-current-frame-out temporal qualification now covers `525`
+samples: the median surface error is stable, while the p95 tail remains too
+large for a free-space claim. The highest-value immediate work is review of the
+`12` temporal jumps and worst prediction tails, then a bounded live-shadow
+queue and separate dynamic-observation layer. Nvblox, raw-scan detectors and
+alternative SLAM remain optional later gates because the current report
+contract does not carry their required ray/timing semantics.

docs/14_LIDAR_DATASET_GATEWAY.md

@@ -49,9 +49,11 @@ Implemented now:
 - a content-addressed `missioncore.k1-local-surface/v1` replay derivative over
   immutable `RAVNOVES00`, with dynamic height/slope/roughness/confidence,
   explicit pose-binding age and conservative observed occupied/unknown policy;
+- leave-current-frame-out next-frame qualification over `525` samples,
+  temporal jump evidence and unverified local-discontinuity candidates;
 - a provider-neutral read-only local-surface view in
   **Парк → Диагностика LiDAR**, synchronized to the five existing
-  RAVNOVES00 scene selectors;
+  RAVNOVES00 scene selectors and a clickable complete-recording timeline;
 - worker storage admission for `D:\NDC_MISSIONCORE\datasets` and
   `/mnt/d/NDC_MISSIONCORE/datasets`;
 - a dedicated, honest dataset catalog in **Полигон → Датасеты**;
@@ -83,7 +85,7 @@ Not implemented:
 - no model training or production promotion;
 - no RELLIS ROS bag admission, continuous synchronized playback or production
   promotion;
-- no rolling-map implementation yet.
+- no ray-cleared free-space or planner-authoritative rolling occupancy map.
 
 ## Product surface boundary