Description
On Tesla Model 3, the car's drive inverter (DI) unexpectedly set DI_cruiseState to UNAVAILABLE while taking a highway curve at ~100 km/h, causing openpilot to disengage. No driver input, no openpilot fault, no panda blocked messages.
Route
https://connect.comma.ai/dffcf1de8723a20f/000000b5--9de0ab8677/777/803
Log analysis
Timeline:
-5.0s angle=0.8° v=122.7 km/h — straight road, engaged
-1.5s angle=-3.5° v=108.6 km/h — entering turn, decelerating
-0.8s angle=-7.6° v=103.9 km/h — deepest part of curve (~7.6° steering)
0.00s DISENGAGE — cruiseState.available=False, accFaulted=False
+0.18s steerTempUnavailable (EPAS went EAC_INHIBITED after cruise dropped)
What was NOT the cause:
steeringDisengage=False — no hands-on-level override, no high angle rate safety faultsteerFaultTemporary=False, steerFaultPermanent=FalsebrakePressed=False, gasPressed=False, steeringPressed=FalsestockAeb=False, stockLkas=False, invalidLkasSetting=FalseaccFaulted=False — not a cruise FAULT, DI went straight to UNAVAILABLE- Panda did not block any messages
Events at disengage:
+0.00s wrongCarMode [noEntry]
+0.00s pcmDisable []
+0.18s steerTempUnavailable [noEntry, softDisable]
The root cause is the car's drive inverter setting cruise to UNAVAILABLE (not FAULT, not STANDBY). The EPAS going to EAC_INHIBITED 0.18s later was a consequence, not the cause.
Hypothesis
Possible ESC/traction control intervention during the curve at highway speed caused the DI to cancel cruise. Would need to check raw CAN on ESP_status (0x145) and ESP_B (0x155) for stability/traction control activation flags around the disengage moment — rlogs are not uploaded for this route.
Description
On Tesla Model 3, the car's drive inverter (DI) unexpectedly set
DI_cruiseStatetoUNAVAILABLEwhile taking a highway curve at ~100 km/h, causing openpilot to disengage. No driver input, no openpilot fault, no panda blocked messages.Route
https://connect.comma.ai/dffcf1de8723a20f/000000b5--9de0ab8677/777/803
Log analysis
Timeline:
What was NOT the cause:
steeringDisengage=False— no hands-on-level override, no high angle rate safety faultsteerFaultTemporary=False,steerFaultPermanent=FalsebrakePressed=False,gasPressed=False,steeringPressed=FalsestockAeb=False,stockLkas=False,invalidLkasSetting=FalseaccFaulted=False— not a cruise FAULT, DI went straight to UNAVAILABLEEvents at disengage:
The root cause is the car's drive inverter setting cruise to UNAVAILABLE (not FAULT, not STANDBY). The EPAS going to EAC_INHIBITED 0.18s later was a consequence, not the cause.
Hypothesis
Possible ESC/traction control intervention during the curve at highway speed caused the DI to cancel cruise. Would need to check raw CAN on
ESP_status(0x145) andESP_B(0x155) for stability/traction control activation flags around the disengage moment — rlogs are not uploaded for this route.