Tesla faked the original 2016 video showing a Model S driving itself without any human intervention, according to former employees interviewed by the New York Times. The allegations were made in a broader story exposing internal divisions at Tesla over its approach to safety in developing and marketing its Autopilot and Full Self-Driving hardware and software.
According to former members of the Autopilot team at Tesla, the car shown in the 2016 video was not operating with only its onboard sensors and software, as the clip implies. In fact, the former Tesla employees allege the route was mapped ahead of time with 3D scanning technology, creating a detailed digital map for the car to follow.
This mapped-ahead-of-time technology is not available to Tesla buyers and the company does not map roads in this way for owners with Autopilot and Full Self-Driving technology on their vehicles. Rather, the company relies on data and video generated by owners driving their cars that is automatically uploaded to the cloud, where useful learnings can be integrated into future software updates that are beamed back to the cars for reference.
More recently and controversially, the company has made its more capable but incomplete Full Self-Driving (FSD) “Beta” software to select owners who rate well on its Safety Score Beta program. This program rewards specific nominally safe driving behaviors, though some users have already figured out how to game it with unsafe driving practices that fool the software into a better score. These customers are not required to have any other qualifications and do not work for Tesla, nor are they restricted as to when or where they can “test” the autonomous driving technology.
The former employees interviewed by the times also allege the Model S shown in the original Autopilot and FSD video hit “a roadside barrier” on Tesla property during filming, halting production while the car was repaired.
The 19 former employees interviewed by the Times were granted anonymity as they feared retaliation from Tesla and its CEO, Elon Musk, who has publicly attacked his detractors in the past, usually on Twitter. Neither Tesla nor Musk responded to multiple requests for comment, according to the Times.
These specific allegations of forgery have the potential to be explosive for Tesla and Musk. Upstart competitor Nikola was accused of and eventually admitted to faking a video of its fuel cell-powered Nikola One semi-truck prototype driving under its own power when in fact it was pushed down a small hill. Some of the footage was later sped-up in editing to give the impression the truck was moving at different speeds.
Soon after the fraud allegations were made in a report by short-seller Hindenburg Research, state and federal investigations were launched into both Nikola and its chairman and founder, Trevor Milton. Milton and Nikola denied the allegations at first, but Milton quickly left the company and was later charged with securities fraud.
“Milton sold a version of Nikola not as it was – an early stage company with a novel idea to commercialize yet-to-be proven products and technology – but rather as a trail-blazing company that had already achieved many groundbreaking and game-changing milestones,” the indictment against Milton reads.
While Milton is accused of other crimes and the extent to which the semi-truck video was doctored was greater, it remains a cautionary tale for Tesla should these new allegations prove true. At their core, both doctored videos present the products as being capable of driving in ways they actually can’t. The Nikola One prototype had no drivetrain to power it, while the Tesla Model S allegedly relied on technology the car wasn’t and still isn’t sold with and omitted a crash that allegedly happened during filming.
The indictment against Milton specifically cites his claims Nikola’s technology was more advanced than it actually was, an accusation that has been leveled against some of Tesla and Musk’s statements around the company’s Full Self-Driving software in the past as well and again by former employees in this Times report.
At the time of publication, no state or Federal regulator has announced an investigation into the Tesla video, which is still available on Tesla’s website and Vimeo channel. Tesla is currently under investigation by the National Highway Traffic Safety Administration over a dozen fatal crashes that have occurred while Tesla vehicles were being operated by Autopilot and Full Self-Driving software.
The Times story links to a video posted on October 20, 2016. The story does not address other videos created by Tesla purporting to show the company’s vehicles driving themselves autonomously with no human intervention. At least three videos showing alleged self-driving were posted in 2016 on October 19, October 20, and November 18. It is unclear whether the other two videos, ostensibly made during the same time period, could also have been doctored, nor is it clear if other videos posted in the intervening years since then may have been doctored.
Using 3D scanning technology to map roadways for use by autonomous and semi-autonomous vehicles is not unusual or uncommon. Indeed, many of Tesla’s competitors in the space, including Ford, Rivian, and General Motors brands Cadillac, Chevrolet, and GMC, employ this technology and restrict their vehicles from being operated on roads that haven’t been mapped. This approach is slower and more expensive as it requires company employees or contractors to map millions of miles of public roads before the hands-free driving technology can be used on them by customers.
Tesla eschews this approach, preferring to rely on data generated by customer cars, which are not restricted from using Autopilot, Autosteer, or Full Self-Driving technologies on any type of road unless the vehicle can’t read the lane lines well enough to engage. Tesla has introduced more semi-autonomous features more quickly than its competitors to date, though the gap is narrowing.
Tesla, and Musk especially, have in the past made pronouncements about the capabilities and advancement of its Autopilot and Full Self-Driving technologies that have not come to pass. Most famously, Musk predicted the company would have “over a million robo taxis on the road” by the end of 2019, if not early 2020. That has not happened, nor did Musk’s follow-up prediction in 2020 that the company would achieve “the functionality necessary for full self-driving by the end of the year.”
At the time of publication, no Tesla robo taxis exist, no fully autonomous vehicle has been sold to the public by Tesla or anyone else, and the few competitors offering commercial autonomous taxi service only do so in extremely limited geographic areas. Tesla’s Autopilot and Full Self-Driving technologies continue to be advanced semi-autonomous driving aids that, according to Tesla, must be monitored at all times by the human driver who is ultimately responsible for the vehicle’s driving and must be ready to take over at any moment if the technology fails or behaves erratically or inappropriately. Though this wording can be found on Tesla’s website and in its vehicles, both the company and Musk have been repeatedly accused of misleading customers about the technology’s capabilities and limitations in other mediums—or at the very least not stepping in to correct overly optimistic readings by the public of those capabilities.
The allegations of video tampering are part of a larger report on Tesla and Musk’s approach towards autonomous driving technology. It reveals numerous internal divisions over the safety of the company’s approach, the controversial decision to rely solely on cameras rather than radar, lidar, and other sensors (like most competitors do), and the grandiose promises Tesla and Musk have made about the technology which may not have always been backed up by the contemporary engineering.
Source: Read Full Article