* modify reset logic
* remove debug statements
* use ecef pos and vel covariances during reset
* reset orientations initialized to 0,0,GPSbearing
* refactor nav fix
* add fake gps observations to control ecef pos and ecef vel std
* replace fake_P with individual fake cov
* set gps mode flag
* add gps invalid flag names
* update refs
* more accurate gps accuracy check + update refs
* set ECEF_POS valid flag to false if in no-gps mode
* modify valid flag for all ecef states before gps
* add calibrated valid when no gps
* update refs
* remove extra whitespace
* add invalid flag to all NED values pre gps
* update refs
* update correct refs
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
* add no gps observations
* use correct ecef_location, orientation vals and stds for nogps states
* remove earth radius obs
* move initial loc to the ocean
* remove unnecessary changes
* update refs
* use canonical language
* filter out when gps signal is flaky
* Update selfdrive/locationd/locationd.cc
Co-authored-by: Willem Melching <willem.melching@gmail.com>
Co-authored-by: Willem Melching <willem.melching@gmail.com>
* use cstring instead of string.h
* use cstdio instead of stdio.h
* remove inttypes.h
* use cstdlib instead of stdlib.h
* use cstdint instead of stdint.h
* #include <cstddef>
* cstdlib
* use cmath
* remove stddef.h
* use cassert
* use csignal
* use ctime
* use cerror
* rebase master
* Add first sanity checks on inputs
* more gps checks
* make sure test values pass sanity checks
* this localizer supports cannonball runs
Co-authored-by: Harald Schafer <harald.the.engineer@gmail.com>
* fix std transform
* 100Hz
* new ref
* no more decimation
* clean up confusing maths
* static typing
* Revert "static typing"
This reverts commit 23d87337de648e629fbd35dd8c04a740bbefca47.
* 100Hz costs more
* move normalization into core
* add quat idxs
* add big eps
* this is not safe in the filter
* more sensible
* updates to rednose
* not tested
* normalize in python too
* update rednose
* nan check
* check for infs too
* all should be finite
* update ref
* rednose pr now in master
Co-authored-by: Harald Schafer <harald.the.engineer@gmail.com>
* C++ Process replay improvements
* revert that change
* create publisher before subscribers
* dont block forever, print warning
* add comment
* create sockets after process init
* try once more to receive the message
* SIMULATION env variable
* print message num when no response
* live_kf to c++
* first locationd code
* Running in process_replay
* locationd handle cam_odo and live_calib
* log event handlers
* working message receiving
* compiling message sending
* correctly sending some messages
* correct receiving and sending
* update ref_commit with some all_alive_and_valid being false, minor fixes
* fix std abs
* linking on device fix
* fix cpu usage test
* generate kf constants and defines
* fix replay test
* replay without acks, cleanup
* operate on bytearray messages
* cleanup
* send msg fix
* small sleep, less flaky test
* remove python locationd
* review feedback
* bump rednose
