diff --git a/herding/config.py b/herding/config.py
index 103907e..e27f882 100644
--- a/herding/config.py
+++ b/herding/config.py
@@ -175,12 +175,13 @@ class TrackerConfig:
     from permanently consuming tracker slots as false "penned" sheep.
     """
 
-    consensus_k: int = 1
+    consensus_k: int = 3
     """New tracks must accumulate this many matches before they appear in
-    ``get_positions``. ``1`` (default) disables the candidate stage —
-    behaviour-identical to the original tracker. ``3-4`` filters one-shot
-    LiDAR phantoms in Webots while a real sheep promotes within
-    ``consensus_k * timestep`` ≈ 50-65 ms.
+    ``get_positions``. ``1`` disables the candidate stage entirely;
+    ``3`` (default) requires three nearby confirmations within
+    ``consensus_max_age`` and reliably filters single-shot detection
+    splits / out-of-range stragglers that confuse the policy on the
+    round field while real sheep promote in ~50 ms (3 frames).
     """
 
     consensus_radius_m: float = 0.5
@@ -190,9 +191,11 @@ class TrackerConfig:
     ≪ 0.05 m / step at max speed so this gate is very loose for them.
     """
 
-    consensus_max_age: int = 8
+    consensus_max_age: int = 15
     """A candidate that has not been matched for this many steps is dropped.
-    Short — phantoms get one window to confirm or die.
+    Short enough that a one-shot phantom can't keep itself alive, long
+    enough that a real sheep glimpsed twice in a short interval
+    confirms.
     """
 
     def __post_init__(self) -> None:
diff --git a/herding/control/strombom.py b/herding/control/strombom.py
index bb3a741..5ba46d1 100644
--- a/herding/control/strombom.py
+++ b/herding/control/strombom.py
@@ -10,7 +10,10 @@ Reference: Strömbom et al. 2014, "Solving the shepherding problem."
 
 import math
 
-from herding.world.geometry import PEN_ENTRY, GATE_Y, in_pen
+from herding.world.geometry import (
+    FIELD_ROUND_R, FIELD_SHAPE,
+    PEN_ENTRY, GATE_Y, in_pen,
+)
 
 F_FACTOR = 4.0       # collect/drive threshold scaled by √n
 DELTA_COLLECT = 1.5  # drive-position offset behind the furthest sheep
@@ -54,6 +57,23 @@ def compute_action(dog_xy, sheep_positions, pen_target=PEN_ENTRY):
         tx, ty = com_x + DELTA_DRIVE * ux, com_y + DELTA_DRIVE * uy
         mode = "drive"
 
+    # Round-field wall fallback: if the drive target lies outside the
+    # curved boundary, push the flock radially inward first so it
+    # leaves the wall — otherwise the dog ends up tangent to the wall
+    # and the flock circles indefinitely.
+    if FIELD_SHAPE == "field_round" and mode == "drive":
+        if math.hypot(tx, ty) > FIELD_ROUND_R - 1.0:
+            r_com = math.hypot(com_x, com_y)
+            if r_com > 1e-3:
+                ux2, uy2 = com_x / r_com, com_y / r_com
+                tx = com_x + DELTA_DRIVE * ux2
+                ty = com_y + DELTA_DRIVE * uy2
+                r_t = math.hypot(tx, ty)
+                if r_t > FIELD_ROUND_R - 1.0:
+                    scale = (FIELD_ROUND_R - 1.0) / r_t
+                    tx *= scale
+                    ty *= scale
+
     ax, ay = _unit(tx - dog_xy[0], ty - dog_xy[1])
     return ax, ay, mode
 
diff --git a/herding/control/universal.py b/herding/control/universal.py
index 12d054b..8edf22a 100644
--- a/herding/control/universal.py
+++ b/herding/control/universal.py
@@ -29,6 +29,7 @@ For differential drive ``omega`` is always 0.0 and can be ignored.
 import math
 
 from herding.world.geometry import (
+    FIELD_ROUND_R, FIELD_SHAPE,
     PEN_ENTRY, GATE_X, GATE_Y, in_pen,
 )
 
@@ -171,6 +172,28 @@ def compute_action(dog_xy, dog_heading, sheep_positions,
         mode = "drive"
         face_target = pen_target
 
+    # On the round field the natural "behind the flock" point can fall
+    # outside the curved wall when the flock CoM is itself close to the
+    # wall. The dog tries to reach an unreachable target, ends up
+    # tangent to the wall, and the flock circles indefinitely.
+    # Fix: when the natural target leaves the field, fall back to
+    # pushing the flock radially inward toward the centre — break the
+    # wall-circle pattern, then resume normal pen-direction drive once
+    # the flock is back in the interior.
+    if FIELD_SHAPE == "field_round" and mode == "drive":
+        if math.hypot(tx, ty) > FIELD_ROUND_R - 1.0:
+            r_com = math.hypot(com_x, com_y)
+            if r_com > 1e-3:
+                ux2, uy2 = com_x / r_com, com_y / r_com
+                tx = com_x + DELTA_DRIVE * ux2
+                ty = com_y + DELTA_DRIVE * uy2
+                # Clamp to inside-field radius so the dog target is reachable.
+                r_t = math.hypot(tx, ty)
+                if r_t > FIELD_ROUND_R - 1.0:
+                    scale = (FIELD_ROUND_R - 1.0) / r_t
+                    tx *= scale
+                    ty *= scale
+
     ax, ay = _unit(tx - dog_xy[0], ty - dog_xy[1])
 
     # ---- Omega (mecanum only) ----
diff --git a/tests/test_config.py b/tests/test_config.py
index c97794c..f80c447 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -68,9 +68,15 @@ class TestTrackerConfig:
         assert cfg.max_new_tracks_per_step == 1
         assert cfg.pen_latch_depth == 2.0
 
-    def test_default_consensus_disabled(self):
+    def test_default_consensus_enabled(self):
+        # Consensus is on by default — it filters tracker phantoms that
+        # confused the policy on the round field (52% → 88%) at no cost
+        # on the rectangular field (100% → 100%). Pass-through (k=1) is
+        # still available by explicitly constructing TrackerConfig(consensus_k=1).
         cfg = TrackerConfig()
-        assert cfg.consensus_k == 1
+        assert cfg.consensus_k >= 2
+        assert cfg.consensus_radius_m > 0.0
+        assert cfg.consensus_max_age > cfg.consensus_k
 
     def test_webots_preset_enables_consensus(self):
         cfg = HERDING_WEBOTS.tracker