foothold_candidates — DRIVE walkterrain op

• データ種: depth → table

• 呼び出し: import fullseye as fs; fs.ledger.foothold_candidates(grid, cell: 'float' = 0.05, window: 'int' = 3, min_score: 'float' = 0.6, min_dist: 'float' = 0.15, max_n: 'int' = 50, extent=None) (実装を直接呼ぶなら import terrain; terrain.foothold_candidates(grid, cell: 'float' = 0.05, window: 'int' = 3, min_score: 'float' = 0.6, min_dist: 'float' = 0.15, max_n: 'int' = 50, extent=None)、台帳から引くなら opsdrive.get("foothold_candidates"))

• GPU: なし(CPU 参照実装のみ)

使い方

Pick discrete safe stepping targets from the terrain.

Scores every cell with foothold_score, keeps local maxima above *min_score*, and greedily suppresses candidates within *min_dist* (world units) of a better one — turning the continuous flatness field into a short list of "here are places you may put a foot". Returns a list of dicts sorted best-first, each with cell (row, col), score, and xy (world) when *extent* is given. The actionable output a foot-placement planner consumes.

参考(サンプルデータ・文献)

• サンプルデータ カタログ(DL URL / ライセンス) — 2-D は skimage.data(BSD/public)+ 合成、3-D は実データ源(Stanford/PDS 等)の DL URL。

• 演算子の来歴・参考文献 — この op 族の元になった研究/手法の出典。

• アルゴリズムの正典(著者・年)と用途は上記ファミリ使い方ガイドに記載。

実行できる例(この op を実際に呼ぶ検証済みサンプル)

• consumer_hillco — py -3.11 examples/consumer_hillco.py

• physical_ai_perception — py -3.11 examples/physical_ai_perception.py

• walking_ops_tour — py -3.11 examples/walking_ops_tour.py

型が繋がる次の op(table を入力に取れる)

course_layout · course_occupancy · course_contains · world_build · world_camera · world_move · lidar_scan · rss_longitudinal_same

同カテゴリ(walkterrain)

elevation_map · fill_gaps · elevation_traversability · foothold_score · ground_surface · ground_plane · detect_obstacles · fuse_elevation

*Provenance: terrain.py — DRIVE operator registry. この per-op ノートは tools/opdocs.py md が自動生成(手編集しない)。*

© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.