Files
StarPilot/selfdrive/test/process_replay
Kacper Rączy 1034dbd37c Online lateral lag learning (#34531)
* Add struct and services

* Proof of concept

* Use vego

* Process replay support

* Fix issues

* Max not min

* Remove print

* Make lag positive

* Points and updates

* Sliding cross corr

* Stuff

* Remove redundant param

* Move to lagd

* Update alert

* Something that works

* Report isEstimated

* Corr field

* Use skimage masked corr

* Fixes

* Move to lat_accel_loc

* Use the delay it in the model

* Mark as executable

* Add 0.2 to initial

* add install line

* Back to 5m window

* Move the import up

* corr window 120

* Show is estimated

* Sort messages

* New impl

* Fix

* Params

* Set initial_lag elsewhere

* Add param

* Rename Param

* Fix type isues

* More blocks

* Masked fft NCC implementation

* Remove package installation

* Use 20hz data and interp the correlation peak

* Move ncc code to separate function

* steering_saturated support

* Constants

* Add recovery buffer

* Fix static

* min_valid_block_count nad liveDelay

* lateralDelayEstimate

* Fix comm issues. Do estimations at 4hz

* Fix

* Display the estimate

* Increase the block size

* Cache every minute

* DEBUG flag

* Add progress to the alert

* Fix crash

* points only in debug

* Fix BlockAverage restoration

* Comment

* Move estimators into estimator directory

* Remove lagd

* Feed only relevant messages

* Rewrite as a set literal

* np.diag with P_initial

* Fix static

* Struct renames

* Update ref commit

* Make it nicer

* Some renames
2025-04-01 04:02:02 +00:00
..
2024-10-02 21:32:56 -07:00
2024-11-13 21:27:23 -08:00
2025-03-03 19:34:49 -08:00

Process replay

Process replay is a regression test designed to identify any changes in the output of a process. This test replays a segment through individual processes and compares the output to a known good replay. Each make is represented in the test with a segment.

If the test fails, make sure that you didn't unintentionally change anything. If there are intentional changes, the reference logs will be updated.

Use test_processes.py to run the test locally. Use FILEREADER_CACHE='1' test_processes.py to cache log files.

Currently the following processes are tested:

  • controlsd
  • radard
  • plannerd
  • calibrationd
  • dmonitoringd
  • locationd
  • paramsd
  • ubloxd
  • torqued

Usage

Usage: test_processes.py [-h] [--whitelist-procs PROCS] [--whitelist-cars CARS] [--blacklist-procs PROCS]
                         [--blacklist-cars CARS] [--ignore-fields FIELDS] [--ignore-msgs MSGS] [--update-refs] [--upload-only]
Regression test to identify changes in a process's output
optional arguments:
  -h, --help            show this help message and exit
  --whitelist-procs PROCS               Whitelist given processes from the test (e.g. controlsd)
  --whitelist-cars WHITELIST_CARS       Whitelist given cars from the test (e.g. HONDA)
  --blacklist-procs BLACKLIST_PROCS     Blacklist given processes from the test (e.g. controlsd)
  --blacklist-cars BLACKLIST_CARS       Blacklist given cars from the test (e.g. HONDA)
  --ignore-fields IGNORE_FIELDS         Extra fields or msgs to ignore (e.g. driverMonitoringState.events)
  --ignore-msgs IGNORE_MSGS             Msgs to ignore (e.g. onroadEvents)
  --update-refs                         Updates reference logs using current commit
  --upload-only                         Skips testing processes and uploads logs from previous test run

Forks

openpilot forks can use this test with their own reference logs, by default test_proccess.py saves logs locally.

To generate new logs:

./test_processes.py

Then, check in the new logs using git-lfs. Make sure to also update the ref_commit file to the current commit.

API

Process replay test suite exposes programmatic APIs for simultaneously running processes or groups of processes on provided logs.

def replay_process_with_name(name: Union[str, Iterable[str]], lr: LogIterable, *args, **kwargs) -> List[capnp._DynamicStructReader]:

def replay_process(
  cfg: Union[ProcessConfig, Iterable[ProcessConfig]], lr: LogIterable, frs: Optional[Dict[str, Any]] = None, 
  fingerprint: Optional[str] = None, return_all_logs: bool = False, custom_params: Optional[Dict[str, Any]] = None, disable_progress: bool = False
) -> List[capnp._DynamicStructReader]:

Example usage:

from openpilot.selfdrive.test.process_replay import replay_process_with_name
from openpilot.tools.lib.logreader import LogReader

lr = LogReader(...)

# provide a name of the process to replay
output_logs = replay_process_with_name('locationd', lr)

# or list of names
output_logs = replay_process_with_name(['ubloxd', 'locationd'], lr)

Supported processes:

  • controlsd
  • radard
  • plannerd
  • calibrationd
  • dmonitoringd
  • locationd
  • paramsd
  • ubloxd
  • torqued
  • modeld
  • dmonitoringmodeld

Certain processes may require an initial state, which is usually supplied within Params and persisting from segment to segment (e.g CalibrationParams, LiveParameters). The custom_params is dictionary used to prepopulate Params with arbitrary values. The get_custom_params_from_lr helper is provided to fetch meaningful values from log files.

from openpilot.selfdrive.test.process_replay import get_custom_params_from_lr

previous_segment_lr = LogReader(...)
current_segment_lr = LogReader(...)

custom_params = get_custom_params_from_lr(previous_segment_lr, 'last')

output_logs = replay_process_with_name('calibrationd', lr, custom_params=custom_params)

Replaying processes that use VisionIPC (e.g. modeld, dmonitoringmodeld) require additional frs dictionary with camera states as keys and FrameReader objects as values.

from openpilot.tools.lib.framereader import FrameReader

frs = {
  'roadCameraState': FrameReader(...),
  'wideRoadCameraState': FrameReader(...),
  'driverCameraState': FrameReader(...),
}

output_logs = replay_process_with_name(['modeld', 'dmonitoringmodeld'], lr, frs=frs)

To capture stdout/stderr of the replayed process, captured_output_store can be provided.

output_store = dict()
# pass dictionary by reference, it will be filled with standard outputs - even if process replay fails
output_logs = replay_process_with_name(['radard', 'plannerd'], lr, captured_output_store=output_store)

# entries with captured output in format { 'out': '...', 'err': '...' } will be added to provided dictionary for each replayed process
print(output_store['radard']['out']) # radard stdout
print(output_store['radard']['err']) # radard stderr