* zero ll patched big model
* probe in a subprocess so usbgpu lock gets released
* compiles
* runs
* num_jobs gets overwritten, use side effect
* poll tg devices
* make sure build crashes on missing gpu
* fine not to rely on Device.default
* seperate tg env for each model runner
* comment
* Revert "seperate tg env for each model runner"
This reverts commit f6470cc4258eaeb3e8e37907ef370871c9af5aa4.
* env is shared, gate on flag
* no fallback warp dev must be set
* build for current device only, unless pc/release
* comment
* list
* listen for plug in
* add icon to status bar, read params on every frame (?)
* log available devices
* try copy out when loading?
* Revert "log available devices"
This reverts commit e8c52a5d59456d4820ecb13b99a6c46ea1386a20.
* Revert "try copy out when loading?"
This reverts commit 518f403aa03faeda1950fe3dbce0d9e4c1584455.
* don't trigger device probe/caching on modeld prepare
* re-export with ll and road edges
* dont cache devices in manager process
* get USBGPU from params
* no usbgpu env
* missed one
* sconscript don't poll
* unconditional env
* always explicitely set devices on input tensors
* set DEV so amd uses right compiler and iface??
* fix flag
* bump tg
* rm xdg_cache_home
* tg don't bump all the way
* missing gmmu=0 at compile time
* dm set dev
* tg backend
* update gitignore
* missing import
* unused imports
* rely on Device.DEFAULT at compile time (already the case bc onnxrunner)
* comments
* dm warp needs DEV set too
* build both smol and big
* misc typos
* set dev at compile time
* don't need
* DEV=CPU when getting metadata, ensure we don't grab gpu lock
* this would also grab lock
* put bool
* warp compile always prepare only
* missed one
* poll ui
* missing here
* don't force usbgpu at build time
* tmp patch fetch_fw
* catch all, follow hardwared patterns
* simpler
* compile make input queues
* revert this
* group this more readable
* rm empty line
* make dummy frame using numpy
* revert compile make input queues
* no compiler at runtime
* cleanup
* fine to rebuild all on change to device node for now
* fix usbgpu_present
* fix sconscript
* no size in header stream decompress
* DEBUG=2
* minimal viable feedback
* egpu gray
* oops
* gotta do this actually
* modeld build only depends on modeld devices
* don't ship onnx to release? or chunk
* don't need
* can only set compiler on dev=
* none device works, will use default
* make linter happy
* chunk agnostic onnx input to compile_modeld
* chunk big onnx
* +x chunker
* fix #!
* and don't ship chunked onnx to release
* firmware now in correct location
* better err on missing onnx/chunk
* SConscript also need to accept chunked onnx
* metadata also need to load maybe chunked
* dedupe cmd
* this needs to be on cpu
* devices are set in the tgflags, we already depend on them
* rebuilding on changed order is fine
* read file chunked can already load either chunked or not
* chunk all big onnx
* less confusing
* unused import
* python device to load onnx bytes
* default device for runners, python for metadata
* why not
* chunked to shm
* speedup slowest
* less setup
* rm som reset logging
* simpliy a lil more
* lil more
* down to 36s
* sleeeeeeep
* finishing touches
* oopsie
* ty fix
---------
Co-authored-by: Comma Device <device@comma.ai>
* intern
* start
* move
* common caching
* use constant for slep
* works
* add gating back
* clean up
* more
* match cache logic
* hate this circular
* not needed since sync
* no need for lock?
* even qt had something like _load_initial_state for tests, keep
* clean up
* clean up
* clean up
* loading json as string works, else it will fail to parse json, catch that and log, and next api call will overwrite
* move over firehose
* clean up
* fix test
* no
* flip
* more
* match qt
* consistent
* clean up
* cmt
* fix test!
* record feedback with LKAS button
* fix alert test
* slightly simplify feedbackd
* "Audio Feedback Saved" upon time expiration or early stop
* earlySend --> earlyStop
* userFlag --> userBookmark
* RecordAudioFeedback param/toggle
* add audioFeedback test
* simplify feedbackd
* send bookmark regardless of toggle, show feedback event with higher priority
* add userBookmark to selfdrived sm
* fix mispelled param name
* default off and move to main
* segmentNum --> blockNum, earlyStop --> lastBlock
* preserve audioFeedback
* get rid of lastBlock and just send bookmark saved at the end
* update raylib side
* update toggle description and add raylib toggle
---------
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
* excessive actuation
* text
* software
* check
* logic
* whoops
* dont want to lose alert unless user interacts with it
* implement
* try x2
* counter
* try to false trigger
* use livepose
* need to check for livePose noise
* cmt
* nl
* add back
* organization
* setVisible and isVisible consecutively don't work
* style
* cant do this sadly
* actually we can!
* clean up
* clean up
* clean up
* need to match torqued, paramsd, lagd, etc. (fix op sim)
* first
* fix
* fix
* this
* real
* time
* time
* more
* more
* fix
* subclass
* fix
* fix
* test
* fix
* ruff
* revert
* default
* json
* more
* test
* test
* d
* better
* better
* test
* one
* test
* space
* add CLEAR_ON_IGNITION_ON
* hardwared is the only place that currently checks all ignition sources. doing it in pandad wouldn't make sense. this okay?
* remember
* only two usages but meh
* manager should probably be the only daemon to manage params
* clean up
* oh this is craze
* Update selfdrive/monitoring/helpers.py
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* together
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* store/send mic audio with toggle
* script to extract audio from logs
* change description and add translation placeholders
* microphone icon
* apply toggle in loggerd
* add legnth and counter
* startFrameIdx counter
* Revert "change description and add translation placeholders"
This reverts commit 7baa1f6de99c6ebe9f9906193da7e83dad79511a.
* send mic data first and then calc
* restore changed description/icon after revert
* adjust fft samples to keep old time window
* remove extract_audio.py since audio is now stored in qcam isntead of rlog
* qt microphone recording icon
* Revert "remove extract_audio.py since audio is now stored in qcam isntead of rlog"
This reverts commit 7a3a75bd8db5376d1e442a3ba931c67550b5f132.
* move extract_audio script and output file by default
* remove length field
* recording indicator swaps sides based on lhd/rhd
* use record icon from comma body
* Update toggle description
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
* update raylib toggle desc cause I did earlier
* microphone --> soundPressure, audioData --> rawAudioData
* cleanup unused var
* update README
* sidebar mic indicator instead of annotated camera
* improve logic readability
* remove startFrameIdx and sequenceNum
* use Q_PROPERTY/setProperty so that update() is actually called on value change
* specify old id for SoundPressure
* fix typo
---------
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
* bad
* Revert "bad"
This reverts commit 6b5475dd90c3a29c00d946c94d726563cbec6179.
* notify param
* doesn't need to live in low level paramcontrol, rename param
* should work
* fix
* disable while engaged
* note
* fix
* just in case
* param is cleared by manager -- this was all to ensure manager got our `started` transition
* clean up
* and that
* rm
* negative better than generic thing
* timer is needed as it's not clean to fix case where you toggle while no ignition -- you can't go onroad + this allows some nice tolerance time for user to switch 2 toggles
* these aren't required or useful
* add to description
* no longer unlive
* allow reset button too
* another pr -- Revert "allow reset button too"
This reverts commit 5d03edddc80d8625ceba5d5178b2781e9d10d9c9.
* rm space from i18n string
* car is powered on
* now alpha long
* fix
* rename param
* thought i did this
* debug
* debug
* more
* more
* moar
* Revert "moar"
This reverts commit e347106246583982e1be206485ad4d8eb944f817.
* Revert "more"
This reverts commit 0746e6d2e400ce1715614c62af84a9680c4d65c1.
* Revert "more"
This reverts commit 0db5a8792c01bb2d687f9152e6ba37fb76d4a453.
* Revert "debug"
This reverts commit 72f235b13727980e4331ff487998ddc7c7f7fb63.
* Revert "debug"
This reverts commit 7871b69f8c1bdffed386c2baac99542fcc50f7eb.
* Online lateral lag learning (#34974)
This reverts commit b4cc9e68d1.
* pad to the best size for fft
* Fix static analysis
* Add typing
* Fix typing
* MAX_LAG
* Calculate cross correlation regardless if the points are valid
* Back to lagd
* Add lagd to process_config
* Lagd in test onroad
* Move lag estimator for lagd
* Remove duplicate entry from test_onroad
* Update process replay
* pre-fill the data
* Update cpu usage
* 25sec window
* Change the meaning of lateralDelayEstimate
* No newline
* Fix typing
* Prefill
* Update ref commit
* Add a unit test
* Fix static issues
* Time limit
* Or timeout
* Use mocker
* Update estimate every time
* empty test
* DT const
* enable RIVIAN again
* Update ref commit
* Update that again
* Improve the tests
* Fix static
* Add masking test
* Increase timeout
* Add liveDelay to selfdrived
* Add liveDelay to selfdrived in process_replay
* Fix block_avg restore after num_blocks
* regen most
* Update bolt
* Update ref commit
* Change the key name
* Add assert
* True weighted average
* LiveParametersV2 key
* Param key
* Fix tests
* Remove old params if error
* Update exception message
* Update the test
* Add test for corrupted byte format in v1
* Rename params_reader => params
* 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
Armand du Parc Locmaria