granular op• Data kinds: image2d × scalar → table
• Call: import fullseye as fs; fs.ledger.repose_angle_heightmap(hm, pitch: 'float', *, ground=None, min_rel_height: 'float' = 0.1, max_rel_height: 'float' = 0.9, bin_deg: 'float' = 0.25, method: 'str' = 'horn', smooth_cells: 'int' = 1) -> 'dict' (to call the implementation directly, import granular; granular.repose_angle_heightmap(hm, pitch: 'float', *, ground=None, min_rel_height: 'float' = 0.1, max_rel_height: 'float' = 0.9, bin_deg: 'float' = 0.25, method: 'str' = 'horn', smooth_cells: 'int' = 1) -> 'dict'; from the registry, opsdrive.get("repose_angle_heightmap"))
Angle of repose φ from a heightmap — mode of the histogram of demops.dem_slope gradients.
Procedure: (1) subtract ground if given (datum correction; otherwise the median of the outermost 1-pixel ring is taken as ground) (2) moving average over smooth_cells (odd) —— mandatory for height maps in which grains are visible: measuring a pile of spheres (grain size 5.3 px) without smoothing gives a gradient mode of 62 degrees (the rims of the spheres); at 2 grain sizes = 11 cells, 24.5 degrees (measured on a MuJoCo pile of spheres). For powder piles with no visible grains, keep it at 1
(3) use only cells in the band from min_rel_height to max_rel_height of the maximum height (removes the rounded foot and blunted top)
(4) count dem_slope(method) in bin_deg steps; φ is the mean of the values falling in the modal bin and its two neighbours (a centroid of bin centres gave a bias of half a bin = 0.125 degrees, measured).
Returns: phi_deg, phi_median_deg, frac_within_1deg (fraction of cells within ±1 degree of the mode —— for a cone ≈ 1, lower with relief), n_cells, hist_centers, hist_counts. Raises ValueError: shape, pitch, options (method only "horn" / "central") / no pile (no cell above the ground) / fewer than 20 cells in the band.
• Sample-data catalog (download URLs / licences) — 2-D uses skimage.data (BSD/public domain) plus synthetic images; 3-D lists download URLs for real data sources (Stanford, PDS, …).
• Operator provenance and references — the sources of the research/methods this op family came from.
• The canonical algorithm (author, year) and its uses are named in the family usage guide above.
• poc_granular_heap_repose — py -3.11 examples/poc_granular_heap_repose.py
table as input)course_layout · course_occupancy · course_contains · world_build · world_camera · world_move · lidar_scan · rss_longitudinal_same
granular)heap_volume_cone · heap_mass · heap_volume_heightmap · beverloo_rate · beverloo_fit · discharge_synth · dispense_mass_from_video · hopper_discharge_rate
*Provenance: granular.py — DRIVE operator registry. This per-op note is generated by tools/opdocs.py md (do not hand-edit).*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.