matt/recon
1
0
Fork 0
mirror of https://github.com/zvx-echo6/recon.git synced 2026-05-20 06:34:40 +02:00
Code Issues Projects Releases Packages Wiki Activity
81 commits 9 branches 1 tag 628 KiB
Commit graph

5 commits

Author SHA1 Message Date
Matt
1a9dfc8f8d feat(offroute): Phase O3b — trail entry index, Valhalla stitching, /api/offroute endpoint 
Phase A: Trail Entry Point Index
- Extract highway endpoints from idaho-latest.osm.pbf using osmium + ogr2ogr
- Store 740,430 entry points in /mnt/nav/navi.db (SQLite with spatial index)
- Entry points by class: service (271k), footway (152k), residential (146k),
  track (111k), path (26k), unclassified (16k), tertiary (9k), secondary (4k),
  primary (4k), bridleway (15)

Phase B: Pathfinder → Valhalla Stitching (router.py)
- OffrouteRouter orchestrates wilderness pathfinding + Valhalla on-network routing
- Queries entry points within 50km (expanding to 100km if needed)
- MCP pathfinder routes to nearest reachable entry point
- Calls Valhalla pedestrian/bicycle/auto costing for on-network segment
- Returns GeoJSON FeatureCollection with wilderness + network + combined segments

Phase C: Flask Endpoint
- POST /api/offroute with start/end coordinates, mode, boundary_mode
- Returns GeoJSON route with per-segment metadata and turn-by-turn maneuvers

Validated: 42.35,-114.30 → Twin Falls downtown
- Wilderness: 0.5km, 9min | Network: 36km, 413min | Total: ~421min
- 21 turn-by-turn instructions, segments connect at entry point

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2026-05-08 13:44:34 +00:00
Matt
3293cb4238 feat(offroute): Phase O3a — trail burn-in, pathfinder seeks trail corridors 
Trail friction REPLACES land cover friction where trails exist:
- Road (value 5): 0.1× friction
- Track (value 15): 0.3× friction
- Foot trail (value 25): 0.5× friction

TrailReader loads /mnt/nav/worldcover/trails.tif rasterized from OSM highways.

Validation shows trail-seeking behavior:
- On-trail travel: 17.3% → 98.7%
- Effort time: 1047 min → 155 min (-85.2%)
- Path travels farther but stays on roads for speed

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2026-05-08 07:26:25 +00:00
Matt
e0eedcedfd feat(offroute): Phase O2c — PAD-US barriers with three-mode boundary respect 
- Add barriers.py: PAD-US raster reader + build_barriers_raster() function
- Rasterize PAD-US Pub_Access=XA (Closed) polygons to CONUS GeoTIFF
- Modify cost.py: boundary_mode parameter (strict/pragmatic/emergency)
  - strict: private land = impassable (np.inf)
  - pragmatic: private land = 5x friction penalty (default)
  - emergency: private land barriers ignored
- Modify prototype.py: three-way comparison output
- Output: padus_barriers.tif at /mnt/nav/worldcover/ (144MB, ~33m resolution)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2026-05-08 06:56:36 +00:00
Matt
26d4bc7478 feat(offroute): Phase O2b — WorldCover friction integration, lake avoidance validated 
- New friction.py: reads WorldCover friction VRT, resamples to match
  elevation grid, provides point sampling for validation
- Modified cost.py: accepts optional friction array, multiplies Tobler
  time cost by friction multiplier, inf for water/nodata (255/0)
- Modified prototype.py: loads friction layer, passes to cost function,
  validates path avoids water cells (friction=255)

Validated on Idaho test bbox:
- Path avoids Murtaugh Lake (no water cells on path)
- Friction along path: min=10, max=20, mean=10.2
- Effort increased 3.4% vs Phase O1 due to friction multipliers

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2026-05-08 06:33:45 +00:00
Matt
f2a0f81580 feat(offroute): Phase O1 foundation — PMTiles decoder, Tobler cost, MCP pathfinder prototype 
- dem.py: Terrarium-encoded PMTiles tile reader with LRU cache
  - Decodes WebP tiles from planet-dem.pmtiles
  - Stitches tiles into numpy elevation grids for arbitrary bboxes
  - Provides pixel-to-latlon coordinate conversion

- cost.py: Tobler off-path hiking cost function
  - speed = 0.6 * 6.0 * exp(-3.5 * |grade + 0.05|) km/h
  - Max slope cutoff: 40 degrees → impassable
  - Returns time-to-traverse (seconds/cell) as cost metric

- prototype.py: Standalone validation on Idaho test bbox
  - 43km × 80km bbox (~17M cells at 14m resolution)
  - scikit-image MCP_Geometric Dijkstra pathfinder
  - Outputs GeoJSON LineString with path metadata
  - Validated: 61.6km path, 21.3 hours effort time

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2026-05-07 23:43:56 +00:00