diff --git a/src/common/steering/base.py b/src/common/steering/base.py
index 40eff1a..cbb365f 100644
--- a/src/common/steering/base.py
+++ b/src/common/steering/base.py
@@ -9,6 +9,13 @@
 
 import numpy as np
 
+from common.vision.line_detector import (
+    ImageParams,
+    LineDetectResult,
+    detect_line,
+    reset_valley_tracker,
+)
+
 
 @dataclass
 class SteeringOutput:
@@ -25,26 +32,78 @@
 class SteeringBase(ABC):
     """操舵量計算の基底クラス
 
-    全ての操舵量計算クラスはこのクラスを継承し，
-    compute メソッドを実装する
+    線検出・レートリミッター・状態管理の共通ロジックを提供し，
+    サブクラスは _compute_from_result で操舵計算のみ実装する
     """
 
-    @abstractmethod
+    def __init__(
+        self,
+        image_params: ImageParams | None = None,
+    ) -> None:
+        self.image_params: ImageParams = (
+            image_params or ImageParams()
+        )
+        self._prev_steer: float = 0.0
+        self._last_result: LineDetectResult | None = None
+
     def compute(
         self, frame: np.ndarray,
     ) -> SteeringOutput:
         """カメラ画像から操舵量を計算する
 
+        線検出 → サブクラスの操舵計算 → レートリミッター
+        の共通フローを実行する
+
         Args:
-            frame: BGR 形式のカメラ画像
+            frame: グレースケールのカメラ画像
 
         Returns:
             計算された操舵量
         """
+        result = detect_line(frame, self.image_params)
+        self._last_result = result
+
+        output = self._compute_from_result(result)
+
+        # レートリミッター
+        max_rate = self._max_steer_rate()
+        delta = output.steer - self._prev_steer
+        delta = max(-max_rate, min(max_rate, delta))
+        output.steer = self._prev_steer + delta
+        self._prev_steer = output.steer
+
+        return output
 
     @abstractmethod
-    def reset(self) -> None:
-        """内部状態をリセットする
+    def _compute_from_result(
+        self, result: LineDetectResult,
+    ) -> SteeringOutput:
+        """線検出結果から操舵量を計算する
 
-        自動操縦の開始時に呼び出される
+        サブクラスで操舵アルゴリズムを実装する．
+        レートリミッターは基底クラスが適用するため，
+        ここでは素の操舵量を返せばよい
+
+        Args:
+            result: 線検出の結果
+
+        Returns:
+            計算された操舵量（レートリミッター適用前）
         """
+
+    @abstractmethod
+    def _max_steer_rate(self) -> float:
+        """1フレームあたりの最大操舵変化量を返す"""
+
+    def reset(self) -> None:
+        """内部状態をリセットする"""
+        self._prev_steer = 0.0
+        self._last_result = None
+        reset_valley_tracker()
+
+    @property
+    def last_detect_result(
+        self,
+    ) -> LineDetectResult | None:
+        """直近の線検出結果を取得する"""
+        return self._last_result
diff --git a/src/common/steering/pd_control.py b/src/common/steering/pd_control.py
index ec31ada..2579b30 100644
--- a/src/common/steering/pd_control.py
+++ b/src/common/steering/pd_control.py
@@ -13,8 +13,6 @@
 from common.vision.line_detector import (
     ImageParams,
     LineDetectResult,
-    detect_line,
-    reset_valley_tracker,
 )
 
 
@@ -46,38 +44,29 @@
         params: PdParams | None = None,
         image_params: ImageParams | None = None,
     ) -> None:
+        super().__init__(image_params)
         self.params: PdParams = params or PdParams()
-        self.image_params: ImageParams = (
-            image_params or ImageParams()
-        )
         self._prev_error: float = 0.0
         self._prev_time: float = 0.0
-        self._prev_steer: float = 0.0
-        self._last_result: LineDetectResult | None = None
 
-    def compute(
-        self, frame: np.ndarray,
+    def _compute_from_result(
+        self, result: LineDetectResult,
     ) -> SteeringOutput:
-        """カメラ画像から PD 制御で操舵量を計算する
+        """PD 制御で操舵量を計算する
 
         Args:
-            frame: グレースケールのカメラ画像
+            result: 線検出の結果
 
         Returns:
             計算された操舵量
         """
-        p = self.params
-
-        # 線検出
-        result = detect_line(frame, self.image_params)
-        self._last_result = result
-
-        # 線が検出できなかった場合は停止
         if not result.detected:
             return SteeringOutput(
                 throttle=0.0, steer=0.0,
             )
 
+        p = self.params
+
         # 位置偏差 + 傾きによる操舵量
         error = (
             p.kp * result.position_error
@@ -100,12 +89,6 @@
         # 操舵量のクランプ
         steer = max(-1.0, min(1.0, steer))
 
-        # レートリミッター（急な操舵変化を制限）
-        delta = steer - self._prev_steer
-        max_delta = p.max_steer_rate
-        delta = max(-max_delta, min(max_delta, delta))
-        steer = self._prev_steer + delta
-
         # 速度制御（曲率連動）
         throttle = (
             p.max_throttle
@@ -116,23 +99,16 @@
         # 状態の更新
         self._prev_error = error
         self._prev_time = now
-        self._prev_steer = steer
 
         return SteeringOutput(
             throttle=throttle, steer=steer,
         )
 
+    def _max_steer_rate(self) -> float:
+        return self.params.max_steer_rate
+
     def reset(self) -> None:
         """内部状態をリセットする"""
+        super().reset()
         self._prev_error = 0.0
         self._prev_time = 0.0
-        self._prev_steer = 0.0
-        self._last_result = None
-        reset_valley_tracker()
-
-    @property
-    def last_detect_result(
-        self,
-    ) -> LineDetectResult | None:
-        """直近の線検出結果を取得する"""
-        return self._last_result
diff --git a/src/common/steering/pursuit_control.py b/src/common/steering/pursuit_control.py
index d576cf3..4e00ac9 100644
--- a/src/common/steering/pursuit_control.py
+++ b/src/common/steering/pursuit_control.py
@@ -14,8 +14,6 @@
 from common.vision.line_detector import (
     ImageParams,
     LineDetectResult,
-    detect_line,
-    reset_valley_tracker,
 )
 
 
@@ -53,37 +51,28 @@
         params: PursuitParams | None = None,
         image_params: ImageParams | None = None,
     ) -> None:
+        super().__init__(image_params)
         self.params: PursuitParams = (
             params or PursuitParams()
         )
-        self.image_params: ImageParams = (
-            image_params or ImageParams()
-        )
-        self._prev_steer: float = 0.0
-        self._last_result: LineDetectResult | None = None
 
-    def compute(
-        self, frame: np.ndarray,
+    def _compute_from_result(
+        self, result: LineDetectResult,
     ) -> SteeringOutput:
-        """カメラ画像から2点パシュートで操舵量を計算する
+        """2点パシュートで操舵量を計算する
 
         Args:
-            frame: グレースケールのカメラ画像
+            result: 線検出の結果
 
         Returns:
             計算された操舵量
         """
-        p = self.params
-
-        # 線検出
-        result = detect_line(frame, self.image_params)
-        self._last_result = result
-
         if not result.detected or result.row_centers is None:
             return SteeringOutput(
                 throttle=0.0, steer=0.0,
             )
 
+        p = self.params
         centers = result.row_centers
 
         # 有効な点（NaN でない行）を抽出
@@ -116,32 +105,14 @@
         steer = p.k_near * near_err + p.k_far * far_err
         steer = max(-1.0, min(1.0, steer))
 
-        # レートリミッター
-        delta = steer - self._prev_steer
-        max_delta = p.max_steer_rate
-        delta = max(-max_delta, min(max_delta, delta))
-        steer = self._prev_steer + delta
-
         # 速度制御（2点の x 差でカーブ度合いを判定）
         curve = abs(near_x - far_x) / center_x
         throttle = p.max_throttle - p.speed_k * curve
         throttle = max(0.0, throttle)
 
-        self._prev_steer = steer
-
         return SteeringOutput(
             throttle=throttle, steer=steer,
         )
 
-    def reset(self) -> None:
-        """内部状態をリセットする"""
-        self._prev_steer = 0.0
-        self._last_result = None
-        reset_valley_tracker()
-
-    @property
-    def last_detect_result(
-        self,
-    ) -> LineDetectResult | None:
-        """直近の線検出結果を取得する"""
-        return self._last_result
+    def _max_steer_rate(self) -> float:
+        return self.params.max_steer_rate
diff --git a/src/common/steering/ts_pd_control.py b/src/common/steering/ts_pd_control.py
index 30944c3..e4f2f41 100644
--- a/src/common/steering/ts_pd_control.py
+++ b/src/common/steering/ts_pd_control.py
@@ -16,8 +16,6 @@
 from common.vision.line_detector import (
     ImageParams,
     LineDetectResult,
-    detect_line,
-    reset_valley_tracker,
 )
 
 
@@ -53,39 +51,30 @@
         params: TsPdParams | None = None,
         image_params: ImageParams | None = None,
     ) -> None:
+        super().__init__(image_params)
         self.params: TsPdParams = (
             params or TsPdParams()
         )
-        self.image_params: ImageParams = (
-            image_params or ImageParams()
-        )
         self._prev_error: float = 0.0
         self._prev_time: float = 0.0
-        self._prev_steer: float = 0.0
-        self._last_result: LineDetectResult | None = None
 
-    def compute(
-        self, frame: np.ndarray,
+    def _compute_from_result(
+        self, result: LineDetectResult,
     ) -> SteeringOutput:
-        """カメラ画像から Theil-Sen PD 制御で操舵量を計算する
+        """Theil-Sen PD 制御で操舵量を計算する
 
         Args:
-            frame: グレースケールのカメラ画像
+            result: 線検出の結果
 
         Returns:
             計算された操舵量
         """
-        p = self.params
-
-        # 線検出
-        result = detect_line(frame, self.image_params)
-        self._last_result = result
-
         if not result.detected or result.row_centers is None:
             return SteeringOutput(
                 throttle=0.0, steer=0.0,
             )
 
+        p = self.params
         centers = result.row_centers
 
         # 有効な点（NaN でない行）を抽出
@@ -127,12 +116,6 @@
         # 操舵量のクランプ
         steer = max(-1.0, min(1.0, steer))
 
-        # レートリミッター
-        delta = steer - self._prev_steer
-        max_delta = p.max_steer_rate
-        delta = max(-max_delta, min(max_delta, delta))
-        steer = self._prev_steer + delta
-
         # 速度制御（傾きベース）
         throttle = p.max_throttle - p.speed_k * abs(slope)
         throttle = max(0.0, throttle)
@@ -140,23 +123,16 @@
         # 状態の更新
         self._prev_error = error
         self._prev_time = now
-        self._prev_steer = steer
 
         return SteeringOutput(
             throttle=throttle, steer=steer,
         )
 
+    def _max_steer_rate(self) -> float:
+        return self.params.max_steer_rate
+
     def reset(self) -> None:
         """内部状態をリセットする"""
+        super().reset()
         self._prev_error = 0.0
         self._prev_time = 0.0
-        self._prev_steer = 0.0
-        self._last_result = None
-        reset_valley_tracker()
-
-    @property
-    def last_detect_result(
-        self,
-    ) -> LineDetectResult | None:
-        """直近の線検出結果を取得する"""
-        return self._last_result
diff --git a/src/common/vision/detectors/valley.py b/src/common/vision/detectors/valley.py
index 0368c5b..e29aa60 100644
--- a/src/common/vision/detectors/valley.py
+++ b/src/common/vision/detectors/valley.py
@@ -99,12 +99,27 @@
         self._frames_lost = 0
 
 
-_valley_tracker = ValleyTracker()
+_valley_tracker: ValleyTracker | None = None
+
+
+def get_valley_tracker() -> ValleyTracker:
+    """モジュール内のデフォルト ValleyTracker を取得する
+
+    初回呼び出し時にインスタンスを生成する
+
+    Returns:
+        ValleyTracker インスタンス
+    """
+    global _valley_tracker
+    if _valley_tracker is None:
+        _valley_tracker = ValleyTracker()
+    return _valley_tracker
 
 
 def reset_valley_tracker() -> None:
     """谷検出の追跡状態をリセットする"""
-    _valley_tracker.reset()
+    if _valley_tracker is not None:
+        _valley_tracker.reset()
 
 
 def _find_row_valley(
@@ -226,9 +241,19 @@
 
 
 def detect_valley(
-    frame: np.ndarray, params: ImageParams,
+    frame: np.ndarray,
+    params: ImageParams,
+    tracker: ValleyTracker | None = None,
 ) -> LineDetectResult:
-    """案D: 谷検出+追跡型"""
+    """案D: 谷検出+追跡型
+
+    Args:
+        frame: グレースケールのカメラ画像
+        params: 二値化パラメータ
+        tracker: 追跡状態（None でモジュール内デフォルトを使用）
+    """
+    if tracker is None:
+        tracker = get_valley_tracker()
     h, w = frame.shape[:2]
 
     # 行ごとにガウシアン平滑化するため画像全体をブラー
@@ -262,7 +287,7 @@
             expected_w = 0.0
 
         # 予測 x 座標
-        predicted_x = _valley_tracker.predict_x(
+        predicted_x = tracker.predict_x(
             float(y),
         )
 
@@ -285,7 +310,7 @@
     )
 
     if len(centers_y) < MIN_FIT_ROWS:
-        coasted = _valley_tracker.coast(
+        coasted = tracker.coast(
             params.valley_coast_frames,
         )
         if coasted is not None:
@@ -308,7 +333,7 @@
         ransac_iter=params.ransac_iter,
     )
     if coeffs is None:
-        coasted = _valley_tracker.coast(
+        coasted = tracker.coast(
             params.valley_coast_frames,
         )
         if coasted is not None:
@@ -317,7 +342,7 @@
         return no_detection(debug_binary)
 
     # EMA で平滑化
-    smoothed = _valley_tracker.update(
+    smoothed = tracker.update(
         coeffs, params.valley_ema_alpha,
     )
 
diff --git a/src/common/vision/line_detector.py b/src/common/vision/line_detector.py
index dbc9ec6..018c10d 100644
--- a/src/common/vision/line_detector.py
+++ b/src/common/vision/line_detector.py
@@ -8,11 +8,17 @@
     reset_valley_tracker, DETECT_METHODS
 """
 
+from __future__ import annotations
+
 from dataclasses import dataclass
+from typing import TYPE_CHECKING
 
 import cv2
 import numpy as np
 
+if TYPE_CHECKING:
+    from collections.abc import Callable
+
 from common import config
 from common.vision.fitting import clean_and_fit
 
@@ -155,6 +161,42 @@
 # ── 公開 API ──────────────────────────────────────
 
 
+def _get_detector_registry() -> dict[
+    str,
+    "Callable[[np.ndarray, ImageParams], LineDetectResult]",
+]:
+    """検出手法の辞書を遅延構築して返す
+
+    Returns:
+        手法識別子と検出関数の辞書
+    """
+    from common.vision.detectors.blackhat import (
+        detect_blackhat,
+    )
+    from common.vision.detectors.current import (
+        detect_current,
+    )
+    from common.vision.detectors.dual_norm import (
+        detect_dual_norm,
+    )
+    from common.vision.detectors.robust import (
+        detect_robust,
+    )
+    from common.vision.detectors.valley import (
+        detect_valley,
+    )
+    return {
+        "current": detect_current,
+        "blackhat": detect_blackhat,
+        "dual_norm": detect_dual_norm,
+        "robust": detect_robust,
+        "valley": detect_valley,
+    }
+
+
+_detector_registry: dict | None = None
+
+
 def detect_line(
     frame: np.ndarray,
     params: ImageParams | None = None,
@@ -170,35 +212,18 @@
     Returns:
         線検出の結果
     """
+    global _detector_registry
+    if _detector_registry is None:
+        _detector_registry = _get_detector_registry()
+
     if params is None:
         params = ImageParams()
 
-    method = params.method
-    if method == "blackhat":
-        from common.vision.detectors.blackhat import (
-            detect_blackhat,
-        )
-        return detect_blackhat(frame, params)
-    if method == "dual_norm":
-        from common.vision.detectors.dual_norm import (
-            detect_dual_norm,
-        )
-        return detect_dual_norm(frame, params)
-    if method == "robust":
-        from common.vision.detectors.robust import (
-            detect_robust,
-        )
-        return detect_robust(frame, params)
-    if method == "valley":
-        from common.vision.detectors.valley import (
-            detect_valley,
-        )
-        return detect_valley(frame, params)
-
-    from common.vision.detectors.current import (
-        detect_current,
+    detector = _detector_registry.get(
+        params.method,
+        _detector_registry["current"],
     )
-    return detect_current(frame, params)
+    return detector(frame, params)
 
 
 def reset_valley_tracker() -> None:
diff --git a/src/pi/main.py b/src/pi/main.py
index 2c61ecc..06892c2 100644
--- a/src/pi/main.py
+++ b/src/pi/main.py
@@ -9,6 +9,7 @@
 
 import dataclasses
 import time
+from typing import Any
 
 from pi.camera.capture import CameraCapture
 from pi.comm.zmq_client import PiZmqClient
@@ -229,6 +230,39 @@
         zmq_client.stop()
 
 
+def _safe_update_dataclass(
+    target: Any,
+    updates: dict[str, Any],
+) -> None:
+    """dataclass のフィールドを型チェック付きで更新する
+
+    Args:
+        target: 更新対象の dataclass インスタンス
+        updates: フィールド名と値の辞書
+    """
+    fields = {
+        f.name: f.type
+        for f in dataclasses.fields(target)
+    }
+    for key, value in updates.items():
+        if key not in fields:
+            continue
+        expected = fields[key]
+        # int フィールドに float が来た場合は変換を許容
+        if expected is int and isinstance(value, float):
+            value = int(value)
+        elif expected is float and isinstance(value, int):
+            value = float(value)
+        elif not isinstance(value, expected):
+            print(
+                f"Pi: パラメータ型エラー "
+                f"{key}: 期待={expected.__name__}, "
+                f"実際={type(value).__name__}"
+            )
+            continue
+        setattr(target, key, value)
+
+
 def _apply_command(
     cmd: dict,
     pd_control: PdControl,
@@ -253,42 +287,36 @@
         for ctrl in [
             pd_control, pursuit_control, ts_pd_control,
         ]:
-            current = ctrl.image_params
-            for key, value in ip.items():
-                if hasattr(current, key):
-                    setattr(current, key, value)
+            _safe_update_dataclass(
+                ctrl.image_params, ip,
+            )
 
     # PD パラメータの更新
     if "pd_params" in cmd:
-        sp = cmd["pd_params"]
-        current = pd_control.params
-        for key, value in sp.items():
-            if hasattr(current, key):
-                setattr(current, key, value)
+        _safe_update_dataclass(
+            pd_control.params, cmd["pd_params"],
+        )
 
     # Pursuit パラメータの更新
     if "pursuit_params" in cmd:
-        pp = cmd["pursuit_params"]
-        current = pursuit_control.params
-        for key, value in pp.items():
-            if hasattr(current, key):
-                setattr(current, key, value)
+        _safe_update_dataclass(
+            pursuit_control.params,
+            cmd["pursuit_params"],
+        )
 
     # Theil-Sen PD パラメータの更新
     if "steering_params" in cmd:
-        sp = cmd["steering_params"]
-        current = ts_pd_control.params
-        for key, value in sp.items():
-            if hasattr(current, key):
-                setattr(current, key, value)
+        _safe_update_dataclass(
+            ts_pd_control.params,
+            cmd["steering_params"],
+        )
 
     # 復帰パラメータの更新
     if "recovery_params" in cmd:
-        rp = cmd["recovery_params"]
-        current = recovery.params
-        for key, value in rp.items():
-            if hasattr(current, key):
-                setattr(current, key, value)
+        _safe_update_dataclass(
+            recovery.params,
+            cmd["recovery_params"],
+        )
 
     # 十字路分類器の遅延読み込み
     if (