locationd: ignore GPS as Kalman input; expand park-cached-output to onroad consumers

locationd: ignore gpsLocation observations entirely (clearpilot_disable_gps
const at the top of handle_gps; falls through to determine_gps_mode's
no-GPS path). gpsd.py keeps publishing real GPS for UI / dashcam / clock
/ night-mode — only locationd ignores it. The previous gpsd.py path was
hard-coding vNED=[0,0,0] while the car was moving 28 m/s, feeding the
Kalman contradictory GPS-vs-IMU velocity observations that propagated into
latcontrol_torque through liveLocationKalman.angularVelocityCalibrated and
caused a real "drift right on straight roads" symptom.

controlsd: short-circuit state_control() while parked. Skips LaC/LoC PID,
MPC, model_v2 reads, lane-change logic — all wasted work when the car
isn't moving. Returns the same enabled=False/latActive=False/longActive=False
CC the original code would have produced, so publish_logs runs unchanged
and card.controls_update keeps the sendcan / tester-present heartbeat
flowing at 100Hz. controlsd CPU dropped from ~59% to ~3% in park.

controlsd: also wire self.FPCC.noLatLaneChange = True/False in the
existing lane-change suppression branch. The Hyundai carcontroller already
reads off frogpilot_variables.no_lat_lane_change for the no-steer signal,
but the UI's distinctive yellow CHANGE_LANE_PATH_COLOR (onroad.cc:901)
reads from frogpilotCarControl.NoLatLaneChange — which nobody was setting,
so the yellow line never appeared during lane-change suppression.

plannerd: skip longitudinal_planner.update + publish + publish_ui_plan
while parked. Downstream controlsd already short-circuits in park, so
longitudinalPlan / uiPlan staleness is fine.

frogpilot_process: same idea — skip frogpilot_planner.update + publish
while parked.

dmonitoringmodeld: mirror the cached-output trick from modeld. Add carState
to the SubMaster, cache the last (model_output, dsp_execution_time) tuple,
and serve it on every frame while gear is in park instead of running DSP
inference on the driver camera.

Together with the existing modeld park-cached-output, the heavy onroad
processing pipeline (modeld → plannerd → controlsd → carcontroller) now
all idles in park and only fires when the car leaves park.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

selfdrive/clearpilot/dashcamd.cc

@@ -12,19 +12,19 @@
  * Trip lifecycle state machine:
  *
  *   WAITING:
- *     - Process starts in this state. Idle.
+ *     - Process starts in this state
  *     - Waits for valid system time (year >= 2024) AND car in drive gear
  *     - Transitions to RECORDING when both conditions met
  *
  *   RECORDING:
  *     - Actively encoding frames, car is in drive
- *     - Gear shift into PARK → close trip immediately → WAITING (idle)
- *     - Ignition off → close trip → WAITING
+ *     - Car leaves drive → start 10-min idle timer → IDLE_TIMEOUT
  *
- *   IDLE_TIMEOUT (deprecated, retained for safety):
- *     - Was used to keep recording across brief drive-thru / fuel stops.
- *     - Now unreachable: drive→park transitions close the trip immediately
- *       and a fresh trip starts on the next drive engagement.
+ *   IDLE_TIMEOUT:
+ *     - Car left drive, still recording with timer running
+ *     - Car re-enters drive → cancel timer → RECORDING
+ *     - Timer expires → close trip → WAITING
+ *     - Ignition off → close trip → WAITING
  *
  *   Graceful shutdown (DashcamShutdown param):
  *     - thermald sets DashcamShutdown="1" before device power-off
@@ -301,11 +301,10 @@ int main(int argc, char *argv[]) {
       }
 
       case RECORDING:
-        // CLEARPILOT: close trip immediately on park (no idle timer). User wants
-        // dashcam idle in park, fresh trip on each drive engagement.
-        if (gear == cereal::CarState::GearShifter::PARK) {
-          LOGW("dashcamd: gear in park, closing trip");
-          close_trip();
+        if (!in_drive) {
+          idle_timer_start = now;
+          state = IDLE_TIMEOUT;
+          LOGW("dashcamd: car left drive, starting 10-min idle timer");
         }
         break;
 

selfdrive/controls/controlsd.py

@@ -681,12 +681,7 @@ class Controls:
       self.LoC.reset(v_pid=CS.vEgo)
       self.frogpilot_variables.no_lat_lane_change = False
       self.FPCC.noLatLaneChange = False
-      # Call LaC.update with active=False so we get the right lac_log subtype
-      # for this car's lateralTuning (torque vs pid vs angle). Internally it
-      # early-returns when active is False — cheap.
-      lp = self.sm['liveParameters']
-      _, _, lac_log = self.LaC.update(False, CS, self.VM, lp, self.steer_limited, 0.0,
-                                       self.sm['liveLocationKalman'], model_data=self.sm['modelV2'])
+      lac_log = log.ControlsState.LateralDebugState.new_message()
       return CC, lac_log
 
     # Update VehicleModel

selfdrive/controls/plannerd.py

@@ -34,21 +34,13 @@ def plannerd_thread():
   while True:
     sm.update()
     if sm.updated['modelV2']:
-      # CLEARPILOT: skip the planning compute while parked, but KEEP publishing
-      # at the normal cadence so consumers' alive flags stay healthy. Skipping
-      # publishes entirely caused longitudinalPlan to go alive=False at
-      # controlsd, which fires a real commIssue the moment we shift out of park.
-      # Stale published values are fine — controlsd's own park short-circuit
-      # ignores the longitudinalPlan content while parked anyway.
-      parked = sm['carState'].gearShifter == car.CarState.GearShifter.park
-      if not parked:
-        longitudinal_planner.update(sm)
+      # CLEARPILOT: skip planning while parked. The downstream consumer (controlsd)
+      # already short-circuits in park, so longitudinalPlan/uiPlan staleness is fine.
+      if sm['carState'].gearShifter == car.CarState.GearShifter.park:
+        continue
+      longitudinal_planner.update(sm)
       longitudinal_planner.publish(sm, pm)
-      # publish_ui_plan reads longitudinal_planner.a_desired_trajectory_full
-      # which is only set inside update(). Skip it while parked — uiPlan is
-      # UI-only, not on controlsd's commIssue check list, so going silent is fine.
-      if not parked:
-        publish_ui_plan(sm, pm, longitudinal_planner)
+      publish_ui_plan(sm, pm, longitudinal_planner)
 
 def main():
   plannerd_thread()

selfdrive/frogpilot/frogpilot_process.py

@@ -85,15 +85,11 @@ def frogpilot_thread():
           frogpilot_planner = FrogPilotPlanner(CP)
           frogpilot_planner.update_frogpilot_params()
 
-      # CLEARPILOT: skip planner compute while parked, but KEEP publishing at
-      # the normal cadence so frogpilotPlan stays alive at consumers. Skipping
-      # publishes entirely caused commIssue ("not_alive: frogpilotPlan") at
-      # controlsd the moment we shifted out of park.
+      # CLEARPILOT: skip planner work while parked.
       parked = sm['carState'].gearShifter == car.CarState.GearShifter.park
-      if sm.updated['modelV2']:
-        if not parked:
-          frogpilot_planner.update(sm['carState'], sm['controlsState'], sm['frogpilotCarControl'], sm['frogpilotNavigation'],
-                                    sm['liveLocationKalman'], sm['modelV2'], sm['radarState'])
+      if sm.updated['modelV2'] and not parked:
+        frogpilot_planner.update(sm['carState'], sm['controlsState'], sm['frogpilotCarControl'], sm['frogpilotNavigation'],
+                                  sm['liveLocationKalman'], sm['modelV2'], sm['radarState'])
         frogpilot_planner.publish(sm, pm)
 
     if not time_validated: