Tesla’s more advanced version of its standard Autopilot feature — its “Full Self-Driving” software — will very soon be available to more drivers. Previously only available to a select number, the FSD Beta v10.0.1 update will be possibly released on Sept. 24 if all goes to plan, so significantly more drivers could be using the software on public roads (depending on one’s driving skills, according to CEO Elon Musk). People who have paid for FSD mode have been promised a button that enables said update.
But the feature comes with some forewarning. A new study by the Massachusetts Institute of Technology suggests that Autopilot may lead to drivers becoming inattentive when administering the advanced driving system.
The model for the study is based upon glance data from 290 human-initiated autopilot disengagement epochs. Essentially, it replicated the observed glance pattern of drivers. MIT’s Alberto Morando, Pnina Gershon, Bruce Mehler, and Bryan Reimer conducted the study by following Tesla Model S and X owners for fractions of a year or more, all based in the greater Boston area.
The results? Visual behaviour amongst drivers is altered before and after Autopilot is disengaged. That means before the feature is switched on/off, drivers look less on the road and pay more attention to “non-driving related areas.” In short, drivers may not be using the feature “as recommended”, becoming less attentive and too relaxed while driving. These shifts in attention are far less common in manual driving.
“This change in behaviour could be caused by a misunderstanding of what the system can do and its limitations, which is reinforced when automation performs relatively well,” write the authors.
It’s important to note that Autopilot and Full Self-Driving mode do not make the car fully autonomous and still require full driver supervision. “Autopilot and Full Self-Driving Capability are intended for use with a fully attentive driver, who has their hands on the wheel and is prepared to take over at any moment,” reads Tesla’s support site. “Full autonomy will be dependent on achieving reliability far in excess of human drivers as demonstrated by billions of miles of experience, as well as regulatory approval, which may take longer in some jurisdictions.”
Though the MIT recognises that Tesla’s safety reports say its current consumers (those who have access to self-driving) are using Autopilot safely, the study urges the implementation of additional systems within the vehicles to monitor driver attention and provide feedback in real time. Tesla’s Autopilot does currently monitor driver engagement with a hands-on-wheel sensing system, but doesn’t consider attention through eye or head regulation. The authors suggest that eye/head tracking is far more linked to paying attention to the road. If applied, the study suggests this could save drivers from inattention and keep them engaged while driving — still hands-free.
Comparatively, Ford and General Motors have hands-free assisted driving systems that use eye/head tracking, with their Blue Cruise and Super Cruise systems, respectively.
A number of high-profile crashes, including a fatal Model X crash in 2018, involving Autopilot have caught the attention of not only investigations such as the MIT study, but also federal forces. Autopilot is currently under investigation by the National Highway Traffic Safety Administration for various safety concerns — including the fact that the feature can be all too easy to trick.
