New Tesla Autopilot Attention Study: Interesting, Not Surprising

–>

The study used MIT Advanced Vehicle Technology to figure out how drivers are influenced by advanced driver-assist systems. In short, it found that people using technology like Tesla Autopilot tend to look away from the road or perform tasks unrelated to driving more often than they would if such technology wasn't in place. The study also calls Tesla Autopilot one of the most capable commercially available systems of its kind, which may be good or bad, depending on you see it.

This isn't rocket science. It makes perfect sense that people would engage a driving assistant and then become somewhat complacent under the right circumstances. However, there are several key details here that must be pointed out. The study says there's not enough evidence to conclude that these systems are unsafe. Moreover, every driver handles such tech differently, and regardless of the "glancing behavior" that the study points out, there's no indication that this means these systems aren't saving lives despite driver beahvior.

The researchers used data from about 500,000 miles of Autopilot driving. They based their model on "glance data from 290 human initiated AP disengagement epochs." Basically, it looks at this glance data among drivers related to the times they chose to disengage the system. Findings were such that "off-road" glances were greater with Autopilot active. Moreover, driving-related glances happened less often while Autopiolot was engaged.

If you asked any seasoned Autopilot driver, they'd probably tell you the opposite. We've heard comments, such as "I pay more attention when Tesla Autopilot or Full Self-Driving Beta are active." While we can see comments like this as someone who may be trying to stick up for Tesla, such a comment could be true even if the data in the study is correct. However, there are likely caveats.

For example, someone new to these driver assist systems is likely to be scared or concerned, so they may actually may more attention, at least for a time. Once they get comfortable with the system, they may glance away, perform other tasks, etc., but only at times that they feel comfortable and confident.

At the same time, once drivers become familiar with and comfortable using a system like Autopilot, while they may glance away more often at times, they could also be more alert and focused at times that they don't feel comfortable and confident about how the system may handle a particular situation. 

As driver assistance systems continue to come to market and improve, it's almost certain that people will feel some sense of confidence such that they can actually allow the car to "assist" them. Isn't that the point here? However, there will also be times when drivers feel anxious that – due to whatever conditions: weather, construction, traffic, edge cases, etc. – they either disengage the system entirely, or they white-knuckle it and pay very close attention so as not to end up in a disaster.

When drivers become over-confident with such systems, there will definitely be accidents, and there already have been several. However, as we know, drivers are prone to taking their eyes off the road too often in any car, especially as smartphones have become more popular and car touch screens are more feature-filled and full of eye candy. In these cases, systems like Autopilot could potentially save lives.

The point here is that while it's interesting to learn that drivers may look away more often when using driver-assistance systems, it's not surprising. We'd need much more data about when and why they looked away, what they were doing, and what happened as a result. Did they look down at their phone and then crash into the back of a stopping car? Perhaps, but such a system should avoid that, right?

What the study doesn't reveal is whether drivers are more attentive when relying on driver-assist system in certain specific scenarios. The study also doesn't get into whether or not, regardless of any changes in driver behavior, cars with systems like Autopilot still save many more lives than cars without such technology.

We encourage you not to read the multiple reports and summaries of this study that just cherry-pick the part about drivers glancing away, since there is no indication that these glances have any impact on increasing accidents or deaths. Instead, head directly to the source of the study and read their findings. It's a good read, and well worth the time.

Again, we strongly encourage you to follow the ScienceDirect source link below and check out the actual study in its entirety. There's a lot to take in. We also encourage you to start a conversation about Tesla Autopilot in our comment section below.

Source: Read Full Article