decide op• Data kinds: table → table
• Call: import fullseye as fs; fs.ledger.check_sequence_score(events, *, maneuver: 'str' = 'lane_change', lead_time: 'float' = 3.0, lead_tol: 'float' = 0.5, max_lead: 'Optional[float]' = None, signal_distance: 'float' = 30.0, distance_tol: 'float' = 3.0, cancel_tol: 'float' = 2.0) -> 'Dict[str, object]' (to call the implementation directly, import drivedecide; drivedecide.check_sequence_score(events, *, maneuver: 'str' = 'lane_change', lead_time: 'float' = 3.0, lead_tol: 'float' = 0.5, max_lead: 'Optional[float]' = None, signal_distance: 'float' = 30.0, distance_tol: 'float' = 3.0, cancel_tol: 'float' = 2.0) -> 'Dict[str, object]'; from the registry, opsdrive.get("check_sequence_score"))
Scores the sequence "safety check → signal → (about 3 s / 30 m) → maneuver → cancel signal" before a lane change or a right/left turn.
events: sequence of {"t": 時刻, "kind": 種類, "s": 道のりの位置(任意)}. Kinds are mirror / head_check (safety check), signal_on / signal_off (signal), start (began changing course / reached the turning point), end (manoeuvre finished). start is required (once).
Rules (sources in SOURCES):
1. There is a signal (signal_missing).
2. There is a safety check before the signal (and before start) (safety_check_missing. Driving manual: "check safety in advance with the rear-view mirror etc. before signalling").
3. Signal timing (signal_timing): for a course change, start − signal ≥ lead_time − lead_tol (driving manual "about 3 seconds before"; the tolerance is an assumption); if max_lead is given, signalling too early is also penalised. For right/left turns and U-turns, start position − signal position ≥ signal_distance − distance_tol (driving manual "30 metres before"; requires s).
4. The signal was not stopped before end (signal_not_continued, Act Article 53 paragraph 1).
5. The signal was stopped within end seconds after cancel_tol (signal_not_cancelled, Act Article 53 paragraph 4; the tolerance is an assumption). If there is no end, 4 and 5 are not checked.
Return value: violations (each {"rule", "detail", "deduction"}), deduction, score = 100 − deduction, ok, lead (seconds, course change) / distance (m, right/left turn).
• 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_driving_decisions — py -3.11 examples/poc_driving_decisions.py
table as input)course_layout · course_occupancy · course_contains · world_build · world_camera · world_move · lidar_scan · rss_longitudinal_same
decide)mirror_reflection_matrix · mirror_virtual_camera · mirror_aim_normal · convex_mirror_fov · mirror_blind_zone · signal_phase_plan · signal_state · predict_amber_onset
*Provenance: drivedecide.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.