VW stops steering after same torque value for > 6 seconds #10
Comments
|
Resolved in 25bcd8f in preparation for upstream. |
jyoung8607
pushed a commit
that referenced
this issue
Dec 17, 2019
b8382bb steerLimitTimer should be car dependent 9a22968 add pa0 temp to ThermalData f6f0f60 Add stock Fcw to carState b608683 no l/r distinction for LDW 555f48d Add ldw alert 8e8b4a4 Remove plusFrame socket in favor of UiLayoutState 3410325 log stock AEB events 2219f2b Add warning about not using cython version of sec_since_boot 8f1a512 for legacy-testing reasons, better to define the used percent instead of avail e86d954 adding low memory event ad23834 remove TODO d0962b3 log mem available and cpu perc in thermald 3b753be Implement error handling and exceptions (#18) a7d5bb7 add explicit dependencies on services.h 1ba6467 fix linter c7d215b Added communityFeatureDisallowed event 492140a Added communityFeature bit detection to CarParams 266a5fe log Panda fault types 347a866 Switch from polling on FIFOs to signal (#12) e25bba7 no need to double build the objects fe43a99 20Hz for radar time step is very standard 2aabf1e Added radar time step to car params e8ae908 Generate capnp for java 57126a2 cereal_shared da655cd Add uptime to health f6a8e39 add test with multiple subscribers 84b3af5 comment out the debugging 4b9c942 added power save state to health packet 66be370 run python unittest in ci 52c6db8 Run scons in CI (#14) 9414615 do need it, but only for arm 2856c37 remove gnustl_shared 7f05ee6 fix apks e3a6bde Revert "no more makefiles" 487fbd0 don't rely on BASEDIR, and add zmq library 223e37a no more makefiles da2ed11 don't link the wrong one fe9fe2a scons builds the python lib now 2f81135 err, it can't build services.h 57b03f8 now we shouldn't need that yaml crap everywhere f8e5327 bridge builds with services.h 2b0cb60 noOutput safety mode is now called silent 83880d5 add msgq tests bcad184 msgq: dont block when fifo does not exists b4b26782 Default to zmq 473e291 fix compilation in docker 30aaadd msgq: try again when no timeout on poll but also no message c4f2ad5 msgq: make sure read_fifos is initalized so we dont close random fds 4e513a85 msgq: dont clean up uninitialized sockerts c008b63 also remove the fifo from disk ef64eb2 MSGQ stability improvements when opening and closing lots of queues e147abc Revert "Revert "deprecate irpwr"" 932dc32 Revert "deprecate irpwr" a684415 disengage ec27e18 capnpc also generated the header files ee52ab9 deprecate irpwr 301c74c8 Merge branch 'master' of github.com:commaai/cereal 6da7d55 add front frame a5944eb add conflate parameter for SubSocket::create ca8df17 Add fault status to health ef4ded0 add conflate support in SubSocket constructor 7fd314a update scons build file 93d814e add saturated flags to indi and lqr logs 50302fe add steeringRateLimited to car.capnp 05e3513 add msgq readme a6759a9 faster make 94b7377 Add struct to log FW version 64ce0b5 add scons build dc9ad18 add debug print statement on SIGINT 4a61269 Merge pull request #10 from commaai/msgq 4873449 use recv one or none after poll a054864 default to msgq fbc4a4c oops bad number 5067cf4 add meta cbd0286 fix export prefix and make shared library world readable c273054 add c exports for jni usage e77f41e zmq already sets the errno correctly 3196cf6 Fix service list path in bridge d35515a add all msgq files, but dont use as default a68a38f Don't delete context from python side only bd46c22 Revert "zmq_ctx_term is blocking" a1fc26b zmq_ctx_term is blocking 0902182 remote address support 21a3536 only delete subsocket when created by same object 34df735 remove extra underscore from __dealloc__ c8748f8 fix internal refs 79b2fbf fixups 23ad256 import messaging and services git-subtree-dir: cereal git-subtree-split: b8382bb
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The VW steering firmware appears to have a stuck-torque safety detection that cancels HCA steering intervention if it receives exactly the same torque value for 6 seconds. Right now, this manifests as a critical LKAS disengagement for OP when "helping" OP around a curve it can't handle on its own. Need to introduce some slight noise at capped points, definitely at 3.00nm, and possibly ditch the 1.00nm cap we're applying on steering-pressed detection, potentially going to turn signal detection instead.
The text was updated successfully, but these errors were encountered: