A new study based on MIT Advanced Vehicle Technology data found that Tesla Autopilot results in a noticeable decrease in driver attention when activated.
We previously reported on MIT Advanced Vehicle Technology’s lab conducting a long-term study on Tesla Autopilot.
Lex Fridman, now famous for his appearances on the Joe Rogan Experience and his own podcast with thought leaders, was previously leading the study, which involves recording Autopilot drives of Model S and Model X owners over nearly 500,000 miles.
They record drivers to understand how they interact with the automated driver-assist features under Tesla’s Autopilot package.
Using data from the study, a group of MIT researchers published a new study called “A model for naturalistic glance behavior around Tesla Autopilot disengagement.”
The study found that drivers tend to look at things non-related to driving more often and for longer periods of time when Autopilot is activated.
They specifically found that drivers would look more at the center screen:
“The model replicates the observed glance pattern across drivers. The model’s components show that off-road glances were longer with AP active than without and that their frequency characteristics changed. Driving-related off-road glances were less frequent with AP active than in manual driving, while non-driving related glances to the down/center-stack areas were the most frequent and the longest (22% of the glances exceeded 2 s). Little difference was found in on-road glance duration.”
This is an issue since the idea behind Autopilot is that it would take away some driving tasks and enable drivers to focus more on staying attentive to the road, which they are required to do since they are still responsible for the vehicle while Autopilot is active.
The study found that drivers looked more at the road after disengaging Autopilot:
“Visual behavior patterns change before and after AP disengagement. Before disengagement, drivers looked less on road and focused more on non-driving related areas compared to after the transition to manual driving. The higher proportion of off-road glances before disengagement to manual driving were not compensated by longer glances ahead.”
Tesla has been releasing quarterly “safety reports”, which they use to claim Tesla vehicles with Autopilot engaged have “close to 10x lower chances of being involved in an accident than the average car“.
However, this interpretation of the data has been disputed.
In most Tesla vehicles equipped with a version of Autopilot (1.0 to 3.0), the Autopilot features are mostly being used for highway driving. The same distinction is found for the “average vehicle”, which NHTSA’s overall crash data in the US is based on.
Accidents are more common on city roads and undivided roads than on the highway.
Obviously, this will depend on the driver. I think Autopilot can certainly help make your driving experience safer.
I’ve personally used it over tens of thousands of miles and I feel like it helps me drive more safely, but I make a point to always pay attention to the road when I use the driver-assistance features.
I think Tesla drivers should be careful not to become too complacent when using the features. Until Tesla can improve the system enough to take responsibility for the driving, it is not a good idea.