A review of Tesla's Autopilot website can be the clearest confirmation that the company has not been able to deliver on Elon Musk's ambitious vision for a self-driving future. "You will be able to call your Tesla from pretty much anywhere," wrote Musk in July 2016. "Once you get up, you will be able to sleep, read, or do anything else [sic] to your destination. "In fact, he predicted that Tesla customers with full self-driving will be able to get their cars to join a riding network to" generate revenue for you while you are at work or on vacation ". In January 201 6, Musk predicted that Tesla cars could drive autonomously the coast to the coast "in ~ 2 years." Needless to say, this has not happened. And after more than two years of incredible vision of Autopilot's future, Tesla's Autopilot page has finally been updated to reflect that reality. The title of the page has been changed from "Full Self-Driving Hardware on All Cars" to "Future Run." An opinion about Tesla's riding network has been removed. In the section "Complete self-driving", a disclaimer is now included that "future use of these features unconditionally depends on achieving reliability that exceeds human drivers as demonstrated by billions of miles of experience." In other words, despite Muskos blustering over the years, Autopilot is still just a driver assistance system. And it will continue to be just a driver assistance system for some time to come. Musk still wants to gradually…
A review of Tesla’s Autopilot website can be the clearest confirmation that the company has not been able to deliver on Elon Musk’s ambitious vision for a self-driving future.
“You will be able to call your Tesla from pretty much anywhere,” wrote Musk in July 2016. “Once you get up, you will be able to sleep, read, or do anything else [sic] to your destination. “In fact, he predicted that Tesla customers with full self-driving will be able to get their cars to join a riding network to” generate revenue for you while you are at work or on vacation “.
In January 201
6, Musk predicted that Tesla cars could drive autonomously the coast to the coast “in ~ 2 years.”
Needless to say, this has not happened. And after more than two years of incredible vision of Autopilot’s future, Tesla’s Autopilot page has finally been updated to reflect that reality.
The title of the page has been changed from “Full Self-Driving Hardware on All Cars” to “Future Run.” An opinion about Tesla’s riding network has been removed. In the section “Complete self-driving”, a disclaimer is now included that “future use of these features unconditionally depends on achieving reliability that exceeds human drivers as demonstrated by billions of miles of experience.”
In other words, despite Muskos blustering over the years, Autopilot is still just a driver assistance system. And it will continue to be just a driver assistance system for some time to come.
Musk still wants to gradually improve the safety of this driver assistance system. Eventually, technology can become so good that it no longer requires human supervision.
But there is reason to doubt that this strategy will work. More importantly, there is reason to worry that it could cause people to be killed.
In 2014, Tesla began delivering the first generation Autopilot hardware, the Society of Automotive Engineers published a five-level taxonomy of autonomous driving systems that predicted driver systems for driving aids (called “level 2” in SAE). jargon) gradually turned into completely autonomous systems that could operate without human supervision (levels 4 and 5).
But the last five years have seen a dramatic shift in industry thinking. Most companies now see driver assistance and full self-driving as distinct markets.
No company has done more to change the industry’s thinking here than Google, whose self-propagation project was spun off as Waymo 2016. Around 2012, Google engineers developed a highway system and let some rank-and-file googlers test it. Drivers warned that the system was not yet completely autonomous, and they were instructed to keep their eyes on the road at all times.
But the self-governing team found that users began to trust the system too quickly. Car cameras showed users “napping, putting on makeup and pussy with their phones.” And it created a great security risk.
“It’s difficult to take over, as they have lost contextual awareness,” said Waymo CEO John Krafcik in 2017.
So Google scrapped plans for a highway assistant and decided to pursue another degree of graduation: a taxi service Originally limited to the Phoenix metropolitan area, Phoenix has large, well-marked streets and snow and ice is rare, so taking a self-drive to Phoenix should be much easier than developing a car with its own driving capability that works in every part of the country and All weather conditions.
This approach also has some other advantages: Self-driving cars benefit from high resolution maps Collecting map data in a single subway area is easier than trying to map the whole world at once.
Self-driving cars also draw benefit from lidar sensors, and the best cost thousands, if not tens of thousands of dollars each It is too expensive for an upgrade to a customer-owned vehicle. But the economy is more viable for a driverless taxi service, because the self-driving system replaces an expensive male taxi driver.
Over the past three years, most other self-driving companies have followed Waymo’s leadership. GM bought a start called Cruise in 2016 and put it to work to develop an autonomous taxi service in San Francisco. Ford made a similar venture to Argo AI 2017. The company is now developing autonomous taxi services in Miami and Washington DC.
Volkswagen and Hyundai are in charge of the Aurora-start with co-founded by Chris Urmson, the former leader of Google’s self-driving project – to develop fully autonomous taxi services. Technical companies such as Uber and Zoox are planning to introduce autonomous taxi services.
Meanwhile, Tesla has pushed forward with its original strategy. For more than two years, Tesla charged customers $ 3,000 or more for a “fully self-driving” package. But progress has been slow. And it has put Tesla in a bond. Leaving the old strategy would probably require repayment of customers who paid for the Full Self-Driving package – which would be both embarrassing and expensive.
Instead, Tesla’s solution has been to move “fully self-driving” target records.  “We already have full self-driving capacity on highways,” said Musk during a January revenue. “So from highway to ramp to highway exit, including passenger cars and moving from one highway to another, there is full self-driving capability.”
Of course, this statement comes with a great asterisk: the driver still needs to monitor the car to make sure it doesn’t crash.
Last week, Tesla announced a rebuilding of autopilot’s pricing structure that reflects this new, more generous definition of full self-driving. Previously, driver features were sold as part of Tesla’s “Enhanced Autopilot” level, which cost $ 5,000. Customers can pay an additional $ 3000 for the “Full Self-Driving” package.
But those who paid for this package got no extra functionality. They waited, yes, “full self-driving” – a car that could drive itself without human supervision.
The new pricing structure completely defines self-driving in different ways. The ability to navigate the motorway changes has, for example, been moved from “Enhanced Autopilot” to the old price structure to “Full Self-Driving” in the new one. Later this year, Teslas with the “Full Self-Driving” package will be able to “recognize and respond to traffic lights and stop signs” and perform “automatic driving on city streets.”
Therefore, Tesla now seems to define “full self-driving” as a system that can handle most road conditions under the supervision of a human driver. Tesla still strives to improve the system adequately – eventually it can work without human supervision. But the new pricing structure makes things less uncomfortable in the meantime, since Tesla can now argue that customers have already received “full self-driving” functions such as the possibility of stopping at stop signs.
Think back to the story of Google’s early beta testers that make makeup or pussy with their phones when they should have monitored Google’s self-driving car prototypes. It is really difficult for a person to pay attention to the road when he rides in a car that mostly drives himself. The better self-driving technology is, the easier it is for a driver’s thoughts to walk and the less likely they are to be ready when intervention is needed.
This dynamic had tragic consequences a year ago when a Uber car hit and killed a pedestrian in Tempe, Arizona. Dashcam video shows that the security driver looks down the knee for several seconds before the crash. Record from Hulu shows that she was streaming a TV show to her phone at that time.
Leading self-driving car companies take a number of precautions to avoid a repetition of this tragedy. Safety drivers receive extensive training before they are allowed behind the wheel. Some companies limit their driver hours. Many companies put two people in each car, one to drive and the other to handle data entry while the driver is awake.
Tesla’s plan is to mainly run a massive driving test project with drivers who use their customers as unpaid security drivers. Drivers do not receive proper training on the dangers of inattentive autopilot use. Tesla does not limit the number of hours people can drive the cars, and the company obviously does not take anyone to sit in the passenger seat.
Tesla takes some important precautions. A Tesla car detects if the driver’s hands are not on the steering wheel, and it challenges a series of escalating warnings – eventually the stop if the driver ignores them. Messages on the screen warn drivers of the dangers of inattentive driving.
There is still reason to doubt that these measures are sufficient for drivers to be involved in the driving task. And this problem will only get worse when Autopilot starts navigating the motorway changes, turning and stopping for stop lights. If your car safely drives you from home from work for 100 consecutive days, it is natural to stop paying attention. If the car makes a serious mistake during the 101th trip, you may not pay enough attention to intervene and prevent a crash.
It just takes a few seconds of inattention to miss a fatal mistake. Data showed that Walter Huang had his hands on the wheel in March 2018 when his Model X steered into a concrete lane divider at 70 miles per hour. Poorly striped paths caused the vehicle to run out of the lane and into the “gore area” – a triangular area of paved road that separated the highway’s travel paths from a departure road. If Huang did not expect model X to make that particular mistake, it would have been easy to assume that this was a stretch of road that did not require its attention.
Musk claims that this test period will be fair shortly because technology becomes much safer than a human driver.
“When will we believe it is safe for full self-drive? Probably at the end of the year,” said Musk during the January interview.
But it seems like another of Musk’s overly optimistic predictions. The varnish on lidar will make this particularly difficult.
Lidar is not a panacea, but one thing is pretty good to make sure that a car does not control directly in large, solid objects such as concrete joints or other vehicles. As recently as October, Autopilot still collided with stuffed cars – something Waymo cars have shown how to avoid for years.
But with lidar and a multi-year head start over Tesla, Waymo has struggled to achieve complete driverless operation in a single metropolitan area. Tesla works to achieve full autonomous operation in a large number of traffic and weather conditions on several continents. It is very difficult to believe that this will happen in 2019.