Author’s Profile: With 20 years of experience in the Internet of Vehicles (IoV) industry, I have likely been in the field longer than many readers have been driving. For the first 13 years, I was rooted in some of the earliest IoV companies in China, participating in the mass production and delivery of the first generation of pre-installed IoV products in the country. For nearly 7 years, I have been involved with a well-known automotive company, leading multiple connected solutions designs and engaging in integrated air-ground-space projects, impacting hundreds of mainstream vehicle models. From suppliers to OEMs, from traditional automakers to new forces, and from hardware development to algorithm optimization, I have witnessed the entire process of China’s IoV technology evolving from ‘crossing the river by feeling the stones’ to ‘leading globally.’ Today, I will leverage my 20 years of experience to help you decode the technical secrets behind the new AECS regulations.
When a vehicle collides at 60 km/h, the AECS (Automotive Emergency Call System, commonly known as eCall) can complete the alert, location, and data upload within 1.2 seconds, which is three times faster than human reaction! In light of the recent unfortunate incident involving Xiaomi, where the system alerted but failed to save lives, today we must use an engineer’s ‘God’s perspective’ to understand the underlying logic of this automotive safety revolution.AECS is not just about ‘calling for help after a crash’!Last year, during a joint debugging of a high-end vehicle in the industry, the team nearly burned the midnight oil in the lab. To ensure that AECS could withstand power loss for 70 minutes (5 minutes of call + 60 minutes of standby + 5 minutes of call), we had to upgrade the battery from 1500mAh to 2200mAh. Don’t underestimate this 700mAh difference; during extreme cold tests at -20 degrees Celsius, a test vehicle caught fire after a collision, and it was the extra power that allowed the system to send the last location 1 minute and 23 seconds before the explosion. Everyone present was sweating—if this data hadn’t been sent out, rescue efforts would have had no way to find the person!Speaking of location, we must praise the unbeatable combination of BeiDou and 5G. Previously, during tests in a mountainous area known as a ‘signal graveyard,’ GPS had a direct error of 15 meters, making it impossible to determine which lane the accident occurred in. After switching to BeiDou, the positioning accuracy improved to 2.3 meters, even without RTK. It was like putting a pin on the map at the accident site. Now, many automakers are secretly using GPS as a backup, while switching the primary to BeiDou—this operation is truly impressive! The country is currently promoting BeiDou independence and priority standards, and it can even apply for government subsidies. Given the recent US-China tariff issues, this seems to become even more significant.The algorithm logic of AECS is more complex than guessing your girlfriend’s thoughts.We set a ‘death threshold’ for collision alerts: only frontal collisions with G-forces exceeding 8g (equivalent to being rear-ended by a large truck) trigger the highest level of rescue. Once, a tester deliberately crashed into a curb, and the system did not alert—because it knew that such minor bumps did not warrant wasting rescue resources.However, most automakers rely on airbag signals to determine whether to trigger alerts, rather than using AECS’s own G-sensor or IMU algorithms for self-assessment, which is another issue altogether.The most troublesome aspect is data security. A few years ago, a supplier tried to save costs by using the AES algorithm for encryption, but it was directly halted by regulatory authorities during cross-border transmission. Now, it is mandatory to use the national secret SM4, with keys automatically changing every 10 minutes. A hacker friend of mine once challenged this and spent three days looking at the encrypted data, finally messaging me saying, ‘This code is insane!’ The Xiaomi incident has exposed the industry’s dirty laundry; it is estimated that 80% of vehicles would be ‘blind’ in such situations.AEB ‘blindness’ is not an isolated case; perhaps automakers are all gambling on probabilities?The incident where Xiaomi crashed into a construction cone exploded in the zero-pressure engineer group. Some have tested 12 mainstream L2-level models, and 80% had recognition rates below 60% for irregular obstacles (cones, water barriers). One new force brand was even more exaggerated, with a pure vision solution mistaking a white guardrail for the sky during heavy rain, resulting in a misjudgment rate of 15%. Of course, there are ‘premium players’: Huawei’s ADS 3.0 uses lidar and millimeter-wave radar to extend the recognition distance for cones to 150 meters, but the cost per vehicle increases by 2000 yuan—this is the truth behind why 100,000 yuan vehicles are often ‘naked.’Speaking of response time, SAE regulations require L2-level systems to leave a 5-second takeover time, but the reality is harsh; the ‘life-saving time’ for humans is being compressed. In a certain new force accident, the time from system alert to collision was only 2 seconds; another automaker faced a direct failure in C-NCAP testing due to a 0.8-second algorithm delay. Now, everyone is working on a ‘distraction prediction system‘; if the steering wheel pressure sensor and DMS camera detect that you have been distracted for more than 8 seconds, the seat vibrates violently and voice alerts bombard you, more intense than my wife urging me to wash the dishes!The ‘compliance war’ among automakers is more thrilling than palace intrigue, and many brands may be secretly cutting costs,with three major technical factions ‘tearing each other apart.’
- Tesla faction: Sticking to pure vision, relying on 1.6 million miles of shadow mode data to train algorithms, but it suffers from ‘cataracts’ under direct sunlight, and the misjudgment rate skyrockets in heavy rain.
- Huawei faction: Material stacking madman, using lidar and high-precision maps to the max; it is said that it can recognize ice cream cones from 200 meters away during demonstrations, but map updates rely on manual review, and users in Xinjiang report, ‘Once you leave the city, you are blind.’
- BYD faction: The most cunning ‘double-faced faction,’ using pure vision for 100,000 yuan models while equipping lidar for 200,000+ yuan models, capturing both cost and market.
How brutal is the supply chain reshuffle?A certain foreign Tier 1 supplier was blacklisted by a new force for not adapting to BeiDou positioning functionality—now, new regulations require support for BeiDou positioning, and suppliers that cannot keep up can only starve.On the contrary, domestic manufacturers are going crazy:NS’s eCall product has captured 28% of the market, and HH’s 5G-V2X module has become a hot seller, even Qualcomm is seeking cooperation.Car owners must read! The following three actions can increase the success rate of AECS saving lives by 90%:
- Perform a ‘system check’ before getting in the car:After starting, long-press the SOS button for 3 seconds (the icon is usually a red phone; some may require 1 second, but 3 seconds can trigger almost all vehicles), and the onboard screen will typically display information like ‘x bars of signal’ and ‘communication module online.’ A friend in the industry encountered a situation last week where the system was offline; upon checking, he found out he hadn’t paid for six months (the automaker is too greedy; this service is charged), almost missing a life-saving opportunity.
- Don’t be lazy in setting emergency contacts:At least bind two people: one family member and one neighbor (to avoid both being unreachable). Remember to enable ‘automatic location sending’; in one case, a car owner was unconscious after an accident, and the system automatically called the property management, reportedly arriving 3 minutes earlier than 120.
- ‘Grab data’ immediately after an accident:Open the automaker’s app, find the ‘black box data’ download; it contains information on speed, steering angle, and braking force before and after the collision—there was a case where a car owner was deemed fully responsible but used this data to prove he had braked hard, successfully overturning the verdict. However, it should be noted that not all apps have this function; you may need to contact the automaker to obtain it.
In the next three years, your car may become a ‘personal bodyguard’; the following black technologies are on the way:
- AEB+AECS linkage: If it detects you are distracted while using your phone, the system will directly reduce the speed to 30 km/h while sending a warning to 110, effectively equipping the car with a ‘anti-risk device.’
- Vehicle-road collaboration to save your life: Streetlight poles in pilot areas in Beijing can directly send signals to vehicles, providing a 1-kilometer advance warning of construction, 8 seconds faster than navigation.
- Saving money while driving is not a dream: A certain insurance company is developing a ‘safety score’; if AECS shows you have zero distractions throughout the year, your premium will be halved—equivalent to getting a free fuel card/electric card every year.
Technology can allow for errors, but don’t gamble your life on probabilities.After countless late nights in the lab, my deepest realization is:No matter how advanced the system, it cannot overcome human psychology of luck.In the Xiaomi incident, the system completed all necessary actions in 1.2 seconds, but the driver’s distraction during those 17 minutes rendered all efforts futile.As engineers, we can strive to achieve 99.99% technical reliability;but as car owners, please always remember:When you hold the steering wheel, you are the last switch for AECS.Next issue prediction (or possibly in a future issue, depending on time):About the backup power for AECS, dissecting the backup power solutions of mainstream automakers.Finally, I want to pose a soul-searching question to everyone:If automakers charged an additional 2000 yuan for AECS, would you be willing to pay?Let’s battle in the comments!