footstep_plan — DRIVE `locomotion` op

• Data kinds: mask → table

• Call: import fullseye as fs; fs.ledger.footstep_plan(mask, cell, start, goal_col, step_max, width_min, width_max, deviation_weight: 'float' = 0.1, extent=None) -> 'dict' (to call the implementation directly, import locomotion; locomotion.footstep_plan(mask, cell, start, goal_col, step_max, width_min, width_max, deviation_weight: 'float' = 0.1, extent=None) -> 'dict'; from the registry, opsdrive.get("footstep_plan"))

• GPU: none (CPU reference implementation only)

Usage

Alternating left/right footstep sequence over a foothold mask (exact shortest path).

> The detailed description below is the original text — the summary and the headings are translated.

足場の候補の mask と、歩幅・歩隔の制約から、左右交互の着地点の列を出す。

格子の約束は terrain.elevation_map と同じ(行 i = y、列 j = x、セル辺 cell [m])。進行方向は +x(列が増える向き)、左足は y が大きい側(行が大きい)。

• 状態 = (いま立っている足のセル, その足が左か右か, 反対の足がゴールを越えているか)。次の一歩は反対の足を、mask が True のセル c に置く:

歩幅 0 ≤ (x_c − x_stance) ≤ step_max [m]、歩隔

width_min ≤ (y_left − y_right) ≤ width_max [m] (左右が入れ替わる交差を許さない)。

• 一歩の費用 = 1 + deviation_weight·[((x_c − x_s − step_max)/step_max)² + ((w − w_mid)/w_half)²](w_mid・w_half は歩隔の幅の中央と半幅、半幅が 0 のときは 2 項目を 0)。第 1 項 1 が歩数を数え、残りは最大の歩幅・中央の歩隔へ寄せる(平地では規則的に交互に並ぶ)。

• ゴール = 最後の 2 歩(左右 1 歩ずつ)の両方が列 goal_col 以上。

• 解法: 状態の上の Dijkstra(費用は非負)= 厳密な最小費用。小さな格子では総当たりの最小と一致する(tests の門)。

start は ((行, 列) の左足, (行, 列) の右足)。どちらの足から踏み出すかも探索が決める。返り値 {"reached", "steps": [{"side": "L"/"R", "cell": (i, j), "xy"(extent があるとき)}], "cost", "n_steps", "expanded"}。届かないときは reached = False・ steps 空・cost = inf(risk_aware_path と同じ約束)。

★roverslip.risk_aware_path は再利用しなかった: あちらの状態は 1 セル、辺は 8 近傍の(8, H, W) の費用で、左右交互・歩幅の窓・歩隔の窓(反対の足との相対位置)を持てない。空の入力(mask が 0 要素)は ValueError で拒否する。

English: exact Dijkstra over (stance cell, side, other-foot-past-goal) states; a step puts the other foot on a True cell within the step-length window [0, step_max] and the step-width window [width_min, width_max] (left foot on the larger-y side). Unreachable goals return reached = False and cost = inf.

References (sample data, literature)

• 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.

Runnable examples (verified samples that actually call this op)

• walking_ops_tour — py -3.11 examples/walking_ops_tour.py

Ops the type connects to (they accept `table` as input)

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

Same category (`locomotion`)

contact_points · com_from_silhouette · support_polygon · com_support_margin · gait_phase · support_line_margin · lipm_capture_point · capture_margin

*Provenance: locomotion.py — DRIVE operator registry. This per-op note is generated by tools/opdocs.py md (do not hand-edit).*

footstep_plan — DRIVE locomotion op