We covered some of the likely human factors implications of automotive assist devices taking control of a driver’s steering in a recent post. The timing was perfect. In conjunction with the Automobile Club of Southern California’s Automotive Research Center, AAA recently conducted a study on lane departure warnings and blind spot monitors (to access the study you need a MA zip code such as 02188). Their findings provide an interest complement to what we discussed last month. We were looking a little further into the future. Their study has more immediate application.
If you are driving and make a jerky movement, would you be OK with the car taking over? If you had fallen asleep? If you had swerved to avoid an animal in the road?
First some background
They define lane departure warnings as a signal that notifies the driver when sensors perceive an unintentional deviation from the current lane due to inattention. The system detects this using a radar-based sensor to watch the lane markers and if you don’t steer enough to stay in the lane (i.e. driver inaction), it assumes the cause it is an unintentional deviation.
They define blind spot warnings as a signal that notifies the driver when sensors perceive an intentional lane switch with when there is an object (usually a vehicle) in the blind spot. The system detects this using a camera that continuously monitors the blind spot and if the steering wheel moves in that direction when an object is in that zone, a warning is presented.
Here are some basic statistics on the prevalence of these devices. I was surprised that they are so common. 50% of 2014 models sold in the US have lane departure detection systems that add steering resistance or slight nudging of the breaks as interventions. This follows what many of the commenters suggested on our last post – subtle intervention that has a real impact on the departure but not enough to lock out the driver.
75% of 2014 models sold in the US have blind spot monitoring systems with subtle warnings such as haptic seat vibration or auditory signals. These don’t actually influence the steering; they are alerts only, without any physical intervention.
It seems that the study defines the difference between these two systems is that lane departures are based on driver inaction whereas blind spot warnings are matched to likely intentional driver actions. Since drivers are constantly adjusting the steering wheel when driving, I am not sure this is the most useful distinction, but it is clearly the threshold at which designers decided whether to implement a physical intervention or just an alert. Countering an intentional action might cause more subjective dissonance and so physical intervention is not seen as the best way to go. At least from a driver preference/technology acceptance point of view.
The article notes a few problems with these systems. Some problems are due to limitations of the technology, but the tech is getting better every year so these should wash out. The blind spot cameras currently have limited range so cars are already in the blind spot before they can be detected, perhaps signaling the driver after it is too late. The cameras also miss motorcycles and other smaller vehicles or pedestrians. The lane departure alerts currently have too many false alarms and missed signals, in part due to faded lane markers.
There are also several potential human factors challenges to the system. At first, the signal will not be well learned and may be more startling than informative. Alerts need to inform the driver that something is wrong and then clue them in as to what it is. Another problem arises after the signal has been well learned. The driver may habituate to the signal and it will not be salient enough to induce any action. The design must navigate (no pun intended) a middle ground between these two. Not too much but not too little signal salience.
Second, just the presence of the system may create a Dunning-Kruger effect (trackback to http://ergonomicsindesign.com/2015/02/confidence-ignorance). Drivers may become overconfident that they are protected from lane departures or steering into a car in their blind spot and therefore adopt a more aggressive driving behavior.
I also wonder if the research center that conducted the study has any HF/E expertise and thought to include these and other human factors principles in their study. It is not clear in the article.
So here are my questions for you. Does this change your opinion compared to our discussion in the previous post about driver assist systems? Do you know anything about the ACSCARC that conducted the study? Or do you have any new thoughts to add?
Image credit: Shuets Udono