First it was the Teslaractioning videos, where owners of the Tesla Model S P85D (and later P90DL) electric car filmed the reactions of unsuspecting passengers as they buried their right foot in the carpet.
Then there was the rush of Tesla Autopilot videos, showing Tesla owners experiencing Tesla’s semi-autonomous Autopilot functionality for the first time. Shortly afterwards came the Tesla Autopilot Fail videos, showing the occasions where still inexperienced Autopilot beta software didn’t do exactly what was expected of it. And as more and more Tesla customers turn Autopilot on, we’ve seen those Tesla Autopilot Fail videos shift from demonstrations of not-so-great driving to reports of actual collisions where Autopilot is accused — at least by the owners of the cars in question — of causing the accident.
The latest case involves a five-day old Tesla Model X P90D which its owner claims the car suddenly accelerated while entering a parking stall, allegedly crossing ‘over 39 feet’ of planters before ultimately crashing into a building, causing significant damage to both the car and the building.
But while the clear allegation from TeslaMotors.com Forum member Puzant is that Tesla’s Autopilot software is to blame for the sudden and violent crash, Tesla is hitting back, claiming logs from the car indicate the driver pushed accelerator to the floor, causing the car to react accordingly and ultimately crash into the building.
Both parties believe the other is wrong, and the spectre of intelligent, malevolent cars has been with us ever since Stephen King released the demonic Plymouth Fury on us in 1983 when he published Christine. So who is right — and how do we deconstruct obvious hyperbole from both skeptics and advocates for Tesla’s Autopilot and autonomous cars in general?
First, it’s worth looking at the details of this incident.
According to Puzant’s posting on the Tesla website, the series of events which led to the unfortunate accident unfolded in the following manner:
“Our 5 day old Tesla X today while entering a parking stall suddenly and unexpectedly accelerated at high speed on its own climbing over 39 feet of planters and crashing into a building.
Alongside the description after being encouraged to do so by other forum members, a series of photographs were uploaded showing the post-crash carnage. They document the damage to the front of the brand-new Model X — its license plate still to arrive — and hole the car made in the building it crashed into. At first glance, they’re pretty scary and seem to back up the claims of Puzant.
But in response to the accident, Tesla told Teslarati that, following examination of associated vehicle logs, the car wasn’t to blame. Instead, it suggested, the driver was the one at fault. Issuing the following statement, the California Automaker refutes any allegations that it was to blame:
“We analyzed the vehicle logs which confirm that this Model X was operating correctly under manual control and was never in Autopilot or cruise control at the time of the incident or in the minutes before.
Data shows that the vehicle was traveling at 6 mph when the accelerator pedal was abruptly increased to 100 percent. Consistent with the driver’s actions, the vehicle applied torque and accelerated as instructed.
Safety is the top priority at Tesla and we engineer and build our cars with this foremost in mind. We are pleased that the driver is ok and ask our customers to exercise safe behavior when using our vehicles.”
Of course, this isn’t the first time that we’ve heard of cars suddenly accelerating and crashing into things. As our friends at GreenCarReports note, past incidents of Unintended Sudden Acceleration have encompassed a variety of models from various years of production, including models by Audi, Toyota and Toyota’s luxury arm Lexus. Usually, the errors were easily explained: Audi’s issue was a result of poor pedal placement while the issues affecting Toyota Prii and certain Lexus models were traced back to either a badly-fitted, oversized carpet or driver error in which accelerator was pressed instead of brake.
But this example — and the other examples involving Autopilot-enabled Tesla electric cars — are the first time we’ve had to contend with privately-owned cars that are technically capable of Level 2 Automated operation on the National Highway Traffic Safety Administration’s five-level Automated Vehicle Classification scale. Since most people aren’t even aware of the existence of the scale and fewer are familiar with the inner working of autonomous or partially autonomous vehicles, it’s led to a lot of confusion, fear and misreporting.
And a giant dose of hyperbole from both sides. So to try and clear things up, we think it’s worth briefly explaining NHTSA’s automated vehicle scale.
At one end (Level 0), a car is fully manual in its operation. There’s no assistance system in place at all, and the driver is fully responsible for every_single_thing the car does. Level 1 meanwhile, covers things like traction control, ABS and rudimentary cruise control. The car takes over basic responsibility to ensure maximum grip is available at all times and can travel at a pre-defined speed without requiring you to put your foot on the throttle. But you’re still responsible for driving the car.
Level 2 is what Tesla’s Autopilot system officially qualifies as, at least in its current beta-test version — and at Level 2 a car can combine two or more safety systems such as adaptive cruise control or lane assist to help the car stay straight and true in its lane. Sophisticated Level 2 cars like Tesla’s Autopilot system can even take over large chunks of your driving duties. But while they may be able to cope with most situations, they are still not truly self-driving. The driver has to retain control at all times, and be ready to jump in at a moment’s notice to take over.
Level 3 continues autonomy further, describing a car that can operate independently for the majority of a trip, giving the driver advanced forewarning that it needs them to assume control of the vehicle. At this level, Drivers can cede control of their car, but are still required to be ready to take over given sufficient warning. Volvo’s DriveMe system is currently a Level 3 system. Tesla’s system, since it doesn’t give forewarning of the need for a driver to take over, is Level 2. And it’s only when you reach Level 4 — in which the car takes over complete responsibility for itself and its driving choices — that you have a truly autonomous vehicle.
Why is this important? Well, for a start, there’s a great deal of hyperbole in the media and Tesla’s fan base when it comes to describing the Model S and Model X as ‘self-driving cars’. And while Tesla itself is careful to note that Autopilot is a ß-software feature that should be supervised at all times, the excitement and giddy glee in which we are collectively consuming the promise of fully autonomous cars means that sometimes we’re failing to appropriately account for dangers present.
Worse still, because we all secretly (or not-so-secretly) want to live in the future, the promise of autonomy means we’re strangely willing to suspend disbelief and go with hype over fact.
Which brings us back to this specific case, and some potential hyperbole on the part of the claimants. While we’re not in a position to comment one way or another as to who is to blame since we don’t have all the information, we can make some simple observations based on what we can see from the supplied images — and what we know of Tesla’s Autopilot operation.
The claim alleges that the Model X in question suddenly drove over the corner of the parking space and climbed “over 39 feet of planters” before crashing into the building. Yet the photographs accompanying the claim show a Model X with its nose less than a foot from the building it crashed into. The rear wheels of the Model X P90D have climbed the curbstone between the parking lot and the landscaped planters, but only just. Its rear still hangs over the tarmac. And while it’s clear the crashed Model X did reverse slightly post-crash (presumably either an action by the driver or the emergency services) the distance between car and building, there’s certainly not 39 feet of planters between the parking space and the building.
The Tesla Model X is 16.5 feet in length. Even if we’re feeling generous and add four feet between the front of the car and the building (and we don’t think it is four feet) the distance between building and parking space cannot be any more than 20 feet at its maximum. A more accurate, if conservative guess would place it nearer to 16 feet.
That’s more than half of the alleged distance and while the acceleration may have begun while the car was 39 feet away from the building, basic math suggests that this would have occurred while the car was still outside the parking space (the average U.S. parking space being between 15 and 20 feet in length). This suggests, but does not prove, that the driver report isn’t entirely accurate. Then again, accident reports rarely are, due to the flooding of our bodies by adrenaline in the post-collision trauma.
That’s not to say we’re siding with Tesla. We don’t have the information to determine who is to fault and we’re not accident investigators. We’re merely noting the point as something we should all remember when reading first-hand accident reports. Trauma does strange things to our brains, while computer data doesn’t (usually) lie.
The most likely scenario we can come up with? That the driver went to apply firm brake pressure to stop the car, and accidentally hit the accelerator instead.
Is Tesla off the hook then? No. While we think it is more likely the car accelerated due to driver error, the way in which Tesla’s Autopilot system operates still needs improvement in order to ensure that drivers really do know when they — or their cars — are in control. Moreover, there are some questions as to why the safety aspects of Tesla’s Autopilot system didn’t kick in to prevent the car from ramming into the building in question.
Additionally, given some of the reports we’ve seen and heard in recent weeks of Tesla Model S and Model X cars crashing due to confusion over Autopilot operation, we’re reiterating our concern that there’s really far too little education over how and when Autopilot should be used. Moreover, we can’t help but wonder if Tesla should have waited until a formal autonomous car framework was in place to ensure that both driver and other road users are aware of and ready for autonomous vehicles — even if they are only Level 2 — before releasing them on the road.
It is something of a PR nightmare. And unless Tesla or someone else steps up to the plate to rectify this situation quickly, we think it won’t take long for safety agencies acting for governments around the world to step in and essentially set back autonomous driving for years or even decades.
Blaming a mechanical ghost in the shell is an easy option. But at the same time, complacency over what the Autopilot system can really do — and what it should really be used for — is just as dangerous. And for that, the automotive world, the legal world, the press and hardened advocates need to ensure that hyperbole has no place in autonomous vehicle discussions.
You can also support us directly as a monthly supporting member by visiting Patreon.com.