openpilot_comma

13264 Commits

123 Branches

82 Tags

5.1 GiB

Tag: Branch: Tree: 097

Commit Graph

Author SHA1 Message Date

Author	SHA1	Message	Date
Shane Smiskol	ee03d13529	card: process that abstracts car interface and CAN (#32380 ) * format card * standalone process * no class member CS, there's no point also can be confusing; what else could be using this? * rename CoS * Update selfdrive/controls/controlsd.py * never works first time :D * canRcvTimeout is bool * hack * add cpu * see what testing closet comes up with * first * some clean up * support passable CI, fix test models * fix startup alert * process replay changes * test_fuzzy * gate carOutput valid on carControl valid * we should publish after we update carOutput * controlsd was using actuatorsOutput from 2 frames ago for torque, not the most up to date * check all checks for carControl in case controlsd dies * log more timestamps * more generic latency logger; needs some clean up latency_logger.py was difficult to understand and modify * card polls on can and carControl to get latest carControl possible * temp try to send earlier * add log * remove latencylogger * no mpld3! * old loop * detect first event * normal send * revert "card polls on can and carControl to get latest carControl possible" how it was is best * sheesh! update should be first * first timestamp * temp comment ( timestamp is slow :( ) * more final ordering, and make polling on/off test repeatable * Received can * new plot timestamps * clean up * no poll * add controllers (draft) * Revert "add controllers (draft)" This reverts commit `e2c3f01b2f`. * fix that * conventions * just use CS * consider controlsd state machine in card: not fully done * hmm it's just becoming controlsd * rm debugging * Revert "hmm it's just becoming controlsd" This reverts commit `534a357ee9`. * Revert "just use CS" This reverts commit `9fa7406f30`. * add vCruise * migrate car state * Revert "migrate car state" This reverts commit `4ae86ca163`. * Revert "add vCruise" This reverts commit `af247a8da4`. * simple state machine in card (doesn't work as is) * Revert "simple state machine in card (doesn't work as is)" This reverts commit `b4af8a9b0a`. * poll carState without conflate * bump * remove state transition * fix * update refs * ignore cumLagMs and don't ignore valid * fix controls mismatch; controlsd used to set alt exp * controlsd_config_callback not needed for card * revert ref temp * update refs * no poll * not builder! * test fix * need to migrate initialized * CC will be a reader * more as_reader! * fix None * init after publish like before - no real difference * controlsd clean up * remove redundant check and check passive for init * stash * flip * migrate missing carOutput for controlsd * Update ref_commit * bump cereal * comment * no class params * no class * Revert "no class" This reverts commit `5499b83c2d`. * add todo * regen and update refs * fix * update refs * and fix that * should be controlsstate * remove controlsState migration CoS.initialized isn't needed yet * fix * flip! * bump * fix that * update refs * fix * if canValid goes false, controlsd would still send * bump * rm diff * need to be very careful with initializing * update refs old-commit-hash: `71f5c441fe`	11 months ago
Adeeb Shihadeh	428f170f59	car: CarController and CarState are always present (#31925 ) * always set * add mock * little more * fix * fix old-commit-hash: `afc96972c8`	1 year ago

Shane Smiskol

ee03d13529

card: process that abstracts car interface and CAN (#32380 )

* format card

* standalone process

* no class member CS, there's no point

also can be confusing; what else could be using this?

* rename CoS

* Update selfdrive/controls/controlsd.py

* never works first time :D

* canRcvTimeout is bool

* hack

* add cpu

* see what testing closet comes up with

* first

* some clean up

* support passable CI, fix test models

* fix startup alert

* process replay changes

* test_fuzzy

* gate carOutput valid on carControl valid

* we should publish after we update carOutput

* controlsd was using actuatorsOutput from 2 frames ago for torque, not the most up to date

* check all checks for carControl in case controlsd dies

* log more timestamps

* more generic latency logger; needs some clean up

latency_logger.py was difficult to understand and modify

* card polls on can and carControl to get latest carControl possible

* temp try to send earlier

* add log

* remove latencylogger

* no mpld3!

* old loop

* detect first event

* normal send

* revert "card polls on can and carControl to get latest carControl possible"

how it was is best

* sheesh! update should be first

* first timestamp

* temp comment ( timestamp is slow :( )

* more final ordering, and make polling on/off test repeatable

* Received can

* new plot timestamps

* clean up

* no poll

* add controllers (draft)

* Revert "add controllers (draft)"

This reverts commit e2c3f01b2f.

* fix that

* conventions

* just use CS

* consider controlsd state machine in card: not fully done

* hmm it's just becoming controlsd

* rm debugging

* Revert "hmm it's just becoming controlsd"

This reverts commit 534a357ee9.

* Revert "just use CS"

This reverts commit 9fa7406f30.

* add vCruise

* migrate car state

* Revert "migrate car state"

This reverts commit 4ae86ca163.

* Revert "add vCruise"

This reverts commit af247a8da4.

* simple state machine in card (doesn't work as is)

* Revert "simple state machine in card (doesn't work as is)"

This reverts commit b4af8a9b0a.

* poll carState without conflate

* bump

* remove state transition

* fix

* update refs

* ignore cumLagMs and don't ignore valid

* fix controls mismatch; controlsd used to set alt exp

* controlsd_config_callback not needed for card

* revert ref temp

* update refs

* no poll

* not builder!

* test fix

* need to migrate initialized

* CC will be a reader

* more as_reader!

* fix None

* init after publish like before - no real difference

* controlsd clean up

* remove redundant check and check passive for init

* stash

* flip

* migrate missing carOutput for controlsd

* Update ref_commit

* bump cereal

* comment

* no class params

* no class

* Revert "no class"

This reverts commit 5499b83c2d.

* add todo

* regen and update refs

* fix

* update refs

* and fix that

* should be controlsstate

* remove controlsState migration

CoS.initialized isn't needed yet

* fix

* flip!

* bump

* fix that

* update refs

* fix

* if canValid goes false, controlsd would still send

* bump

* rm diff

* need to be very careful with initializing

* update refs
old-commit-hash: 71f5c441fe

11 months ago

Adeeb Shihadeh

428f170f59

car: CarController and CarState are always present (#31925 )

* always set

* add mock

* little more

* fix

* fix
old-commit-hash: afc96972c8

1 year ago

2 Commits (097)