Phantom braking still an issue
- Thread starter Mr Miserable
- Start date
Does this occur when using auto pilot? It doesn't brake when you are in manual control does it??
planehazza
Member
WannabeOwner
Well-Known Member
2020.8 has been reported to have been causing emergency braking in situations which were never an issue previously - such as oncoming traffic on a bend on a country road. Supposedly that version has been pulled, but if installed on a car it will still be there until the next version arrives (historically cars with "Pulled versions" have been prioritised for next version), or Tesla issues a regression version (which they have done in the past, but extremely rarely)
Not only is phantom braking worse with the new software update (in AP) but I’m having ‘corrective steering’ warnings on roads I have happily driven since November with no change to my driving habits and style. Here’s hoping for a new iteration soon
SimonT
Member
I haven’t noticed phantom breaking for some time.
I do get the corrective steering warning quite a bit.
I do get the corrective steering warning quite a bit.
WannabeOwner
Well-Known Member
Sorry, it was "corrective steering" that I meant to mention wrt 2020.8, rather than emergency braking.
VanillaAir_UK
Supporting Member
Phantom braking worse due to software version or due to lighting?
There seems to be a general consensus that some phantom braking events are caused by strong shadows from say overhead structures even though one would have thought that radar might have played a bigger part in this. But I can confirm a phantom brake event (early jan) heading towards a bridge in bright sunlight that I had passed beneath without incident many times before and after. So there does seem to be something in this.
I assume when people talk about latest version they are meaning 2020.08.x rather than 2020.12 which last time I looked was not on wide spread release?
exlatccatsa
Supporting Member
On my way back from picking up my M3 from Edinburgh on Monday, I experienced it. barely 10 miles into the drive home.
Just like air in the fuel line causing an ICE to cut out momentarily.
I associated it with the change in road surface light to dark, but didn't have it again even though I was ready for it. Thank goodness for threads like these so I understood what was going on.
Its embarrassing/ dangerous for drivers following though.
Just like air in the fuel line causing an ICE to cut out momentarily.
I associated it with the change in road surface light to dark, but didn't have it again even though I was ready for it. Thank goodness for threads like these so I understood what was going on.
Its embarrassing/ dangerous for drivers following though.
planehazza
Member
I might start getting into the habit of hitting the dash cam button when it happens.
My phantom braking has been largely due to other vehicles. The other day I was overtaking a parked forestry lorry giving it plenty of space and no traffic coming in the opposite direction in quite a bright stretch of road and the car braked fiercely. I have very narrow roads around me and cars crossing my front at an angle from left to right as I approach a left hand bend is probably the key time for this event. I just prepare to override it at known bends
In case anyone not seen:
Phantom breaking
My take is that there are at least two or three distinct behaviours here. One related to overtaking large vehicles, another driving under overhead signs / tunnels (which imo purely by coincidence often cast a shadow on the road in front of car (see @verygreen 's tweet) and third changed behaviour with oncoming traffic / congested roadway / parked cars / pedestrians etc.
The first two may be related and relate to secondary radar reflections coming of the high vertical surfaces then off the road surface. The increase in interventions on smaller roads is likely part of an attempt to tighten up object identification.
The first two may be unlinked as I have noticed quite a different level of severity between the two. The couple of dozen or so real 'phantom brake' events (hard braking slowing at least by 20 mph) I have had are (were - proportion changed significantly with 8.1 to 50/50 or even more on small roads) roughly 20% on small / City roads and 80% on dual carriageways / motorways. Of the 80% I would give 45% due to incorrect GPS speed data esp near on / off ramps / slip roads, 50% possible reflection including minor slowing passing trucks) and the rest misc.
I think the passing truck dip happens so often and is mild in comparison so I wonder if it is a mis-trigger of the 'not passing traffic stream too fast' feature?
Based on @verygreen 's tweet, the car maintains a kind of probability / liklihood matrix and in some cases it suddenly becomes certain that there is an object when there is a big enough cluster of reflections.
Phantom breaking
My take is that there are at least two or three distinct behaviours here. One related to overtaking large vehicles, another driving under overhead signs / tunnels (which imo purely by coincidence often cast a shadow on the road in front of car (see @verygreen 's tweet) and third changed behaviour with oncoming traffic / congested roadway / parked cars / pedestrians etc.
The first two may be related and relate to secondary radar reflections coming of the high vertical surfaces then off the road surface. The increase in interventions on smaller roads is likely part of an attempt to tighten up object identification.
The first two may be unlinked as I have noticed quite a different level of severity between the two. The couple of dozen or so real 'phantom brake' events (hard braking slowing at least by 20 mph) I have had are (were - proportion changed significantly with 8.1 to 50/50 or even more on small roads) roughly 20% on small / City roads and 80% on dual carriageways / motorways. Of the 80% I would give 45% due to incorrect GPS speed data esp near on / off ramps / slip roads, 50% possible reflection including minor slowing passing trucks) and the rest misc.
I think the passing truck dip happens so often and is mild in comparison so I wonder if it is a mis-trigger of the 'not passing traffic stream too fast' feature?
Based on @verygreen 's tweet, the car maintains a kind of probability / liklihood matrix and in some cases it suddenly becomes certain that there is an object when there is a big enough cluster of reflections.
Last edited:
I feel that to complete the picture of what is going on my last 2 experiences with heavy braking needs to be included. i.e. on both occasions the Max speed was adjusted down from 70 mph to 30 mph and both happened as the motorway I was driving on passed over roundabouts beneath the motorway - the weather was cloudy, so I believe the inaccurate GPS signal triggered it somehow, believing the car was entering the roundabouts beneath motorway at 50+ mph.
VanillaAir_UK
Supporting Member
We have suffered several of those and I hope will be easy to dramatically reduce occurrences of *if* (speculation that it does not) it used a bit more contextual awareness of the events leading up to the event.
On one of our occurrences, we were on NOA with a route that remained on the motorway. The car did not indicate to leave the motorway, no steering adjustments were made to the path, yes the car still believed that we had suddenly left the motorway and was heading at 70 towards an overhead roundabout.
On other third party reported occasions, the car slams on because suddenly, instead of travelling at highway speed, it thinks that it is travelling on a different road that crosses the road that it has been travelling on. Not an intersection, a completely different road.
Lots more examples can be found of the car over reacting to a moment in time incident rather than taking into account the context it is in. Hopefully a relatively easy fix.
As for shadows vs radar, I have come to the conclusion that the car is more sensitive to false alarms in lighting conditions where there are sudden changes in lighting conditions, ie strong sunlight and shadows especially overhead structures.
The jury is still out for me on that.
This is a strong factor for me too, and I see many 'shadow / lighting' scenarios where the car doesn't brake, but many do coincide with a large, flat structure generally high up like a gantry sign that could cause reflections as per @verygreen. I'm sure that's not the only cause and I certainly agree with @mccgregor that GPS related includes the car misinterpreting which road it's on.
Slightly different point... AP / NoA seems to have its own data that is not posted on signs but that deals with slip roads and a few other situations such that the AP speed self adjusts down then back up if it thinks you are on one of these. In the US, the road layout convention is that at an off ramp, if you are on the inside lane, you MUST be turning off so you have to slow down.
@mccgregor, keep an eye if you see the same slowing as you pass under other roads if you are in the middle or right hand lanes. So far I have not.
In France it was very predictable on the autoroutes. At every slip road if I was on the inside lane the car adjusted speed as though I must be turning. It held that speed right to the end of the whole junction where the on ramp slip road had rejoined.
@mccgregor, keep an eye if you see the same slowing as you pass under other roads if you are in the middle or right hand lanes. So far I have not.
In France it was very predictable on the autoroutes. At every slip road if I was on the inside lane the car adjusted speed as though I must be turning. It held that speed right to the end of the whole junction where the on ramp slip road had rejoined.
Drew57
Member
I drove on Autopilot on the A14 through the new A1/Cambridge section yesterday & the car started to brake heavily from 72 mph at one point, virtually an emergency stop. Luckily nothing was following & I took control almost instantly. The sky was grey & I wasn't driving under anything so no shadows, straight piece of road, one car a long way in front & nothing visible in the rear view mirror.
Google maps navigation showed the new road correctly so I can only assume that the car somehow picked up a former road (or a tiny bit from a construction contraflow) in its path at that point. On other roads I occasionally notice that the speed symbol on the screen is different to the actual road.
It begs the question - what does Tesla actually use to determine a roads' speed?
This was part of a mammoth journey from Chester via M6/A14 to assist my Father (outdoors) near Bury-St-Edmunds, supercharging at Elveden for the return - around 11 hours driving including the supercharger stop.
At one point near the Gravelly Hill Interchange when using Autopilot/Autosteer in heavy traffic the moving graphic display locked up & Autopilot disengaged for a couple of minutes - looked like the car lost sight of the lane lines so my car and those around remained static on the screen for about 1/2 mile even after the traffic spread out and we moved on. It then corrected, I switched back to Autopilot & everything was fine. I hope that's a one-off.
.....It's made me much more wary about these sort of things happening when using Autopilot.
(if anyone here driving a blue LR, followed by a blue M3P passed a white LR, M6 near Coventry yesterday morning, that was me!)
Google maps navigation showed the new road correctly so I can only assume that the car somehow picked up a former road (or a tiny bit from a construction contraflow) in its path at that point. On other roads I occasionally notice that the speed symbol on the screen is different to the actual road.
It begs the question - what does Tesla actually use to determine a roads' speed?
This was part of a mammoth journey from Chester via M6/A14 to assist my Father (outdoors) near Bury-St-Edmunds, supercharging at Elveden for the return - around 11 hours driving including the supercharger stop.
At one point near the Gravelly Hill Interchange when using Autopilot/Autosteer in heavy traffic the moving graphic display locked up & Autopilot disengaged for a couple of minutes - looked like the car lost sight of the lane lines so my car and those around remained static on the screen for about 1/2 mile even after the traffic spread out and we moved on. It then corrected, I switched back to Autopilot & everything was fine. I hope that's a one-off.
.....It's made me much more wary about these sort of things happening when using Autopilot.
(if anyone here driving a blue LR, followed by a blue M3P passed a white LR, M6 near Coventry yesterday morning, that was me!)
Last edited:
And so say all of us!
It seems like a mix of poor map based data possibly supplemented by some of Tesla's own ideas on the matter.... at or near certain junction types or going from one motorway to another.
Similar threads
