I think I need to correct an impression that may have been given by Peter's answer.
A full ATP system is likely to reduce the likelihood of a SPAD because it will be braking a train as it approaches a signal at red, BUT it will NOT stop the SPAD. There has to be a tolerance to compensate for the odometry uncertainty, so whereas the speed will be severely limited it has to allow the driver to drive right up to the signal and hence even past it. The ATP is designed to ensure that the "release speed" at which it permits the driver to SPAD is low enough that the ATP can be sure that it will stop the train by the defined safe place (generally the end of the overlap)- i.e. once it is certain that the train really must have passed the signal, it can then put on the emergency brakes and have enough distance left that it stop the train from the limited speed. Therefore ATP should reduce the consequence of the SPAD to zero (always remembering that the things that actually stop the train are the brakes and the adhesion between wheel and rail, either of which can fail but are typically not the responsibility of the signal engineer.)
The same is true in ETCS.
Indeed it is fundamentally the same in the CBTC metro world. However since often implemented without any lineside signals, then there is a view that no "overlaps" are provided- the truth is though that there is
1. a place where the Automatic Train Operation attempts to stop the train and
2. a separate place a little further on which is the place where the ATP guarantees to stop the train.
If you provide a large number of position references on the approach to the stopping position, the odometry uncertainty can be kept very low and this helps minimise the distance needed between these two locations, but there is always a minimum distance between where it "should stop" and where it "must stop". Such a safety margin may not always be very obviously there, but it does exist. You can call it an overlap beyond the stopping position or you can say that the train should stop a little prior to the full distance, but this is just playing with words.
A full ATP system is likely to reduce the likelihood of a SPAD because it will be braking a train as it approaches a signal at red, BUT it will NOT stop the SPAD. There has to be a tolerance to compensate for the odometry uncertainty, so whereas the speed will be severely limited it has to allow the driver to drive right up to the signal and hence even past it. The ATP is designed to ensure that the "release speed" at which it permits the driver to SPAD is low enough that the ATP can be sure that it will stop the train by the defined safe place (generally the end of the overlap)- i.e. once it is certain that the train really must have passed the signal, it can then put on the emergency brakes and have enough distance left that it stop the train from the limited speed. Therefore ATP should reduce the consequence of the SPAD to zero (always remembering that the things that actually stop the train are the brakes and the adhesion between wheel and rail, either of which can fail but are typically not the responsibility of the signal engineer.)
The same is true in ETCS.
Indeed it is fundamentally the same in the CBTC metro world. However since often implemented without any lineside signals, then there is a view that no "overlaps" are provided- the truth is though that there is
1. a place where the Automatic Train Operation attempts to stop the train and
2. a separate place a little further on which is the place where the ATP guarantees to stop the train.
If you provide a large number of position references on the approach to the stopping position, the odometry uncertainty can be kept very low and this helps minimise the distance needed between these two locations, but there is always a minimum distance between where it "should stop" and where it "must stop". Such a safety margin may not always be very obviously there, but it does exist. You can call it an overlap beyond the stopping position or you can say that the train should stop a little prior to the full distance, but this is just playing with words.
PJW