A new study questions the efficacy of car accident-avoidance systems, but it’s possible that a simple smartwatch might be part of the solution.
Volvo
Today’s connected cars try to deliver a lot of safety features, such as using radar and video cameras to detect an imminent accident and sometimes take over braking and steering control in an effort to avert that accident. At the very least, the cars will issue increasingly loud and bright visual and audio alerts to get the driver to take action to avert the accident.
But a new University of Missouri College of Engineering study calls into question whether such systems work and, far more frightening, whether they might indeed be counterproductive and make drivers less safe. Yet it’s possible that a simple smartwatch might be part of the solution. Ah, the joys of IoT and connected computing never end.
The problems with these systems — sometimes called advanced driver assistance systems (ADAS) — come down to two issues: alert fatigue (where the driver gets so annoyed by the frequent and seemingly pointless alerts that the driver either ignores them or turns them off), and distractions that come at the worst possible instant.
Although alert fatigue can make the system unable to improve safety, it’s the distraction problem that can make the experience literally less safe. According to Jung Hyup Kim, an assistant professor of industrial and manufacturing systems engineering in the University of Missouri College of Engineering and a primary author of the study, there are two possible driver states at the moment of alert that a serious accident is imminent: The driver either already knows about the situation and is actively working at this instant to avoid the accident; or the driver doesn’t already know about it.
If the driver happens to not have noticed anything yet, then the alerts can be lifesaving and critically helpful. But if the driver has already detected the problem and is actively trying to solve it, an insistent alert at the instant can distract the driver at the worst possible time. In theory, that distraction alert might even cause a brief delay in response, which itself could prove fatal.
What does Kim suggest doing about this? He is proposing a system that will try to determine if the driver is already aware of the problem and is actively engaged in protective measures. Kim’s proposed system will feature video cameras in the dashboard, which will watch the eyes and look for changes suggesting that the driver is already in panic mode, as well as a wrist-based sensor that will track changes in muscle tightening. That wrist-based sensor would likely be a smartwatch and quite possibly integrated into an existing popular smartwatch.
To get more technical and specific, let’s look at how the report describes these monitoring mechanisms: The system will use “physiological measures to advance our understanding of the relations between human biometric signals and physical reactions, such as pupillary response, electroencephalography, electromyography signals, etc. The dilation of pupil could reveal the upcoming choice of the human about half-a-second before the actions in the dynamic decision-making process. Also, other studies have supported that the patterns of pupillary response could reflect human perceptual decisions. Therefore, in this study, we assumed the pupil dilation changes could be one of the indicators of physical responses once drivers give a physical reaction to the collision warning. Also, electromyography (EMG) is an electrodiagnostic medicine technique for evaluating and recording the electrical activity produced by skeletal muscles. As it quantifies muscle force levels, it has been widely used to classify human postures, with decoding user intent and indicating patients’ physical movement in healthcare. So it was introduced as another indicator of physical movement in this study. Therefore, the research objective is to predict drivers’ upcoming physical reactions to collision avoidance warnings by using the ratio of pupil diameter changes and EMG responses as two physiological measures.”
In a phone interview with Computerworld, Kim compared the distraction problem to a husband driving a car with his wife a passenger when a pedestrian runs out into the street. This prompts the wife to shout at her husband, ostensibly to flag the pedestrian problem. “The wife screaming might make you more frightened,” Kim said, and therefore less able to maneuver the car quickly to avoid hitting the pedestrian.
The EMG sensor is looking for “a strong muscle constriction that we might be able to detect” whereas eye movement could be tracked by either a dashboard camera or having the driver wearing an eye-tracking device, such as $20,000 eye-tracking glasses.
I would argue that the glasses are a bad idea. The only way for this approach to work — and it would be up to car manufacturers to do it — is to make the system seamless, piggybacking on existing systems, such as a smartwatch the driver might already be wearing. A dashboard camera would be invisible to the user. And Apple, for example, would probably love the idea of embedding EMG sensors in a future Apple Watch so that the latest version of the watch would be pushed by Toyota or Ford or whoever.
If this works, it would sharply reduce unnecessary alerts (“unnecessary” being defined as trying to tell the driver something the driver already knows), which should reduce both alert fatigue and distractions during emergencies. It would be a welcome improvement. Now if I can only get my car to stop beeping when I have the audacity to put a bag of groceries on the front passenger seat without then connecting the seatbelt.
Credit: Evan Schuman, Computerworld
