Jia Haonan From the Passenger Temple
AI4Auto | Intelligent Car Reference
Tesla’s latest annual report is full of words:
“I can’t understand why anyone still doesn’t buy Tesla”…
Why?
The report cites data from various sources to prove that Tesla is both safe and environmentally friendly, for example, the accident rate of Tesla’s autonomous driving is 8 times lower than that of human drivers:
In terms of active safety, Tesla also directly compared with competitors:
Not long ago, a major accident happened in Beijing due to a Tesla owner’s misstepping on the accelerator, leading to a collision with 4 cars at a speed of 170 kilometers per hour. At first, the owner claimed that “the brake failed”, but later admitted his own operation error.
In this report, Tesla also disclosed for the first time the active safety measures for human errors.
In addition, internal combustion engine enthusiasts complain about electric vehicle fires, and Tesla also presents data to prove that the probability of its own products catching on fire is 11 times lower than that of traditional cars:
How did Tesla come to this conclusion, and how does Tesla’s definition of vehicle safety differ from that of traditional automakers?
Why is Autopilot 8 times stronger than humans?
The so-called “8 times stronger” is based on the latest statistical data from NHTSA (National Highway Traffic Safety Administration) in 2021.
First, how many miles on average cause an accident when Autopilot is activated.
Then, the average number of miles recorded for a vertical and non-Autopilot Tesla owner with an accident, as well as a horizontal comparison with the same data from the NHTSA in the United States.
Tesla stated that when Autopilot (including automatic assisted driving and active safety functions) is activated, one accident is recorded per 4.54 million miles (7.26 million kilometers).
For Tesla owners who have not activated Autopilot and have also turned off active safety functions, an average of one accident occurs every 1.3 million miles (2.08 million kilometers).## Translation
In 2021, data recorded by the US NHTSA showed that on average, there was one traffic accident per 550,000 miles (880,000 kilometers) across the US.
The so-called 8x multiple is obtained by comparing Tesla Autopilot’s accident rate when activated with the national average accident rate.
The statistics do not include accidents involving FSD, which remains an assisted driving level of Autopilot. Tesla also considers the statistics to be conservative.
Tesla claims to be rigorous in its statistics, counting every accident where Autopilot is disengaged within five seconds prior to the collision and every accident where the airbag is deployed, even if it is a rear-end collision with no fault assigned.
Furthermore, NHSTA’s accident statistics do not take minor accidents into account, and Tesla believes at least half the accidents are missed.
This means that the numerator may be even higher, and the reliability of Autopilot may be much more than 8 times higher than the national average.
These statistics are sourced from Tesla’s explanations on three aspects.
First, there are the active safety features of the vehicles, such as automatic emergency braking, lane departure warning, front and side collision warning, obstacle recognition, and so on.
We are all familiar with AEB, ACC, and collision warning, but Tesla has disclosed a unique active safety feature for the first time: the pedal misapplication remedy mechanism.
If the camera recognizes a nearby obstacle in front and the driver quickly presses the accelerator pedal, the system judges it as a misapplication and immediately cuts off the electric motor torque output to ensure safety.
Tesla claims that this feature can prevent hundreds of accidents every month.
The so-called ‘brake failure’ accidents involving Tesla might also be correspondingly reduced.
Behind these active safety features is the data-driven feature iteration that Tesla emphasizes.
Currently, over 2 million vehicles are on the road worldwide and constantly providing empirical data. This scale is much larger than any other autonomous driving company.
This allows Tesla to discover complex scenarios that are difficult to cover in ordinary autonomous driving tests and regulatory standards and provide targeted solutions.After accumulating enough data, the Autopilot system improves its accuracy in accident anticipation and can tighten the seatbelt and trigger the airbag before a collision occurs.
Tesla also presented passive safety data, such as the lower center of gravity due to the lack of a front-mounted engine, making Tesla similar to a mid-engine gasoline vehicle. This not only increases the overall flexibility and maneuverability of the car, but also maximizes the reduction in rotational kinetic energy along the horizontal axis and improves the stability of the vehicle. Moreover, Tesla has strengthened the body structure specifically for common collision types.
As for its effectiveness, Tesla stated that both Model 3 and Model Y received a 5-star rating in all USNCAP categories.
Regarding safety, Tesla not only listed third-party data as proof but also explained some of its own understanding in the report. For example, Tesla previously faced controversy over airbag problems, and it explained why sometimes airbags do not deploy in some accidents. For instance, in a common “side overlap” collision, previous sensor designs could only detect forward collisions and were insufficient in detecting side impacts.
Therefore, Tesla later improved the layout and type of safety airbag sensors, enabling Tesla to detect nearly all types of collisions. In terms of battery safety, gasoline car enthusiasts used to criticize electric cars for their potential to catch fire, which eventually became an invisible danger that discouraged many users.
However, Tesla presented data in the report, proving that although it cannot entirely avoid fire accidents, its fire accident rate per 1 billion miles is 11 times lower than the comprehensive data of all cars in the United States. Moreover, Tesla emphasized the effectiveness of its safety configuration, providing better assurance to users with even the same active safety configuration, as sensor coverage and underlying computing capabilities vary greatly among vehicles.Tesla’s active safety features are ensured by eight cameras, a neural network, and a complex electromechanical system, and its AEB performance is more than 45% better than common visual+radar fusion schemes. The company has received high scores in tests conducted by the IIHS (Insurance Institute for Highway Safety), Euro NCAP (European New Car Assessment Program), ANCAP (Australasian New Car Assessment Program) and other third-party organizations.
In summary, Tesla’s official statement is that their vehicles focus on safety, reasonable design, and conscientious materials. However, what Tesla does not explicitly mention is that its drivers’ continuous contribution of road data, with each accident enriching the Tesla database, actually makes them unpaid test drivers and safety experimental materials for Tesla.
So, has today’s Tesla Autopilot become safer than yesterday’s? This is now a matter of detailed statistical analysis, as presented by international media outlet insideevs.
All data used for analysis originated from official Tesla statistics. The meaning of the data is that accidents are reported every XX miles, so the higher the position on the y-axis, the better the performance.
Regarding Tesla’s safety controversy, the opinions from both sides are often highly polarized. The focus of the debate tends to center around incidents, such as voluntary recalls, spontaneous combustion, overly or insufficiently cautious brake response and other operational glitches in Autopilot.
What does Tesla’s official position on this issue, then? In this latest report, Tesla’s choice of words offers some room for interpretation.
Tesla claims that the media often reports only the electric vehicle fires in order to grab attention, and never mentions traditional combustion engine fires and spontaneous combustion.
In fact, as early as February, Musk posted three tweets in a row complaining about the “double standard” treatment Tesla received from the media.
“More than 1 million gasoline vehicle fires each year cause thousands of deaths, but none of them make headlines like a Tesla car fire. Why should there be a double standard? This is a problem.”
Do you think Musk is trying to exonerate Tesla’s operational glitches, or is the media really treating Tesla unfairly?## One more thing
In a Tesla report, the company disclosed information about its work and compensation.
In 2021, Tesla received over 3 million job applications, and according to the 2021 Universum survey, Tesla and SpaceX were the top two preferred employers for graduates in STEM fields in North America.
So what kind of compensation does Tesla offer?
Tesla did not disclose specific salary amounts but introduced a stock-based compensation system for employees.
Assuming an employee received 320 shares of Tesla stock as a reward in 2018, they can sell 20 shares per quarter for 16 quarters (4 years).
The employee’s actual profits come from the difference in stock prices when selling the shares. Based on 2018 Tesla stock prices, by the beginning of 2022, the employee can earn an average of $14,113 (RMB 94,000) per quarter in addition to their salary.
Plus, employees can purchase additional stock at a discount through Tesla’s stock purchasing plan.
That means earning an additional $400,000 a year at least. Are you tempted?
Finally, here is the Tesla China recruitment page for those interested:
https://www.tesla.com/careers/search/?country=CN
Tesla’s annual report link:
https://www.tesla.com/ns_videos/2021-tesla-impact-report.pdf
— End —
This article is a translation by ChatGPT of a Chinese report from 42HOW. If you have any questions about it, please email bd@42how.com.