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>
Two fixes that reinstate the pre-revert defenses against the "TAKE CONTROL
IMMEDIATELY / Communication Issue" banner that fires when self-driving
on the baseline modelrevert stack:
calibrationd: publish valid based on calStatus == calibrated, not
sm.all_checks(). Original gate cascaded upstream freq glitches into
liveCalibration.valid=False, which kept locationd.filterInitialized
False, which fed garbage into paramsd, which corrupted steerRatio
(erratic steering). "valid" here is a question about convergence, not
input freshness.
controlsd: narrow the commIssue trigger to genuine comm failures —
not_alive OR can_rcv_timeout. The `not sm.all_checks()` branch also
picked up valid=False, but paramsd / torqued / plannerd / frogpilot_planner
/ dmonitoringd all propagate their sm.all_checks() into msg.valid via
a polling-pattern artifact (freq_ok inside poll='...' subscribers
tracks gaps between drain bursts rather than the publish rate), so
the whole stack flaps valid and trips the banner during normal
driving. Content and rate are fine; just the flag.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Narrow re-introduction of the power-save mode: when carState reports
gearShifter == park, serve the last rendered model_output instead of
running GPU inference. First cycle (or before carState is flowing)
still runs one real inference so we have something to cache and serve.
Out-of-park returns to per-frame inference immediately.
Avoids the pitfall of the earlier variable-rate path: this doesn't
change the publish rate on modelV2/cameraOdometry (we still publish
every frame), so downstream freq_ok / valid checks in calibrationd /
locationd / paramsd stay happy. Only the GPU inference is skipped.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
controlsd:
- stop writing the removed 'no_lat_lane_change' persistent param
(key isn't in the whitelist anymore). Write to
frogpilot_variables.no_lat_lane_change instead so the carcontroller
read still works and the lane-change-disengage feature is preserved.
- initialize self.driving_gear = False in __init__; it's set at the
end of step() in update_frogpilot_variables but clearpilot_state_control
reads it on the first iteration before that's run.
modeld:
- remove the remaining CLEARPILOT variable-rate / standby code path
that was baked into the initial-import commit (the one our first
commit captured). Constant 20fps, no ModelStandby/ModelStandbyTs
writes, no params_memory reads for 'no_lat_lane_change' (which
isn't registered anymore).
- drop now-unused locals (model_standby, last_standby_ts_write,
params_memory).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Re-adds the non-self-drive controlsd integrations that the UI and
memory-param pipeline need:
- import SpeedState from clearpilot.speed_logic
- self.speed_state, speed_state_frame, was_driving_gear init
- subscribe to gpsLocation (ignored in alive/freq_ok/valid gates — OK if
the GPS daemon isn't publishing)
- clearpilot_state_control:
- auto-reset ScreenDisplayMode 3→0 on park→drive transition
- ~2Hz speed-state update driving the speed-limit sign and cruise
over/under warning sign via ClearpilotCruiseWarning /
ClearpilotSpeedLimitDisplay memory params
The debug-button (LFA) ScreenDisplayMode cycling already lived in the
reverted baseline (it was in the first commit), so it's preserved.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Starting point for rebuilding self-drive from a known-good baseline.
Reverts the following to their state at f46339c:
- selfdrive/modeld/modeld.py (constant 20fps, no variable-rate / standby skip)
- selfdrive/modeld/dmonitoringmodeld.py (no carState sub, no standstill skip)
- selfdrive/controls/controlsd.py (no parked-cycle skip, no FPCC hoisting, no MDPS split)
- selfdrive/controls/lib/longitudinal_planner.py
- selfdrive/locationd/calibrationd.py (valid = sm.all_checks again)
- selfdrive/locationd/paramsd.py
- selfdrive/locationd/torqued.py
All non-self-drive features (thermald fan control, speed limit controller,
cruise warning signs, UI state transitions, GPS fixes, ClearPilot menu,
dashcamd, telemetry, etc.) remain as-is on this branch — only the 4 core
self-drive processes are reverted.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-19 13:26:36 -05:00
4 changed files with 8 additions and 197 deletions
- [sunnypilot — raising torque for heavier vehicles](https://community.sunnypilot.ai/t/raising-the-torque-for-heavier-vehicles/862)
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
brianhansonxyz