Tesla, the major electric vehicle manufacturer, has launched its FSD (Full Self-Driving) system in countries such as the US, Canada, China, and Australia. With a cumulative 4.7 billion miles of data, the company claims that it is 10 times safer than human drivers. What will the future hold? What challenges will it face? This article provides a detailed overview. (Background: Tesla's FSD successfully navigates a vegetable market in China; is Taiwan next?) (Supplementary background: Tesla achieves 'factory automated driving to the dock'; Musk: AI heralds the era of unmanned FSD) Tesla officially opened its Full Self-Driving (FSD) in Australia and New Zealand this week, allowing owners to enable it directly through an OTA update without needing to return to the dealership. This marks the first time both countries can use the full version of FSD, marking a milestone for Tesla's expansion in autonomous driving in the Asia-Pacific region, while also sparking lively discussions on safety, adoption rates, and regulations. This article will start with definitions and characteristics, exploring the current usage status, global approval situation, future development plans, challenges faced, and the multifaceted impacts on society, economy, and safety. What is FSD? Core features Full Self-Driving (FSD) stands for 'Full Self-Driving Assistance', an advanced driving assistance system developed by Tesla. According to the standards of the Society of Automotive Engineers (SAE), FSD currently falls under Level 2 of autonomous driving technology, which means that although the system can handle most driving tasks, the driver must remain attentive and ready to take control at any time. The current version of FSD uses cameras, radar, and ultrasonic sensors on the vehicle, combined with artificial intelligence (AI) to achieve intelligent navigation, steering, lane changes, parking, and summon functions. It can automatically plan routes and handle complex traffic situations, such as city streets, roundabouts, and highway merging. Unlike traditional Autopilot, FSD focuses more on end-to-end (E2E) AI learning, processing visual inputs using neural networks rather than relying on traditional maps or LiDAR. This makes the system more like a human driver and able to adapt to unknown environments, but it also requires continuous training to improve accuracy. What are SAE autonomous driving levels? Level 0: No Automation System Capability: No autonomous driving functions at all. The vehicle may have some warning systems (e.g., blind spot detection alerts), but the system itself does not intervene or control the vehicle's steering, acceleration, or braking. Driving Role: The driver must perform all driving-related tasks at all times. Common examples: Traditional old cars, vehicles without cruise control. Level 1: Driver Assistance System Capability: The system can assist with one driving task, such as controlling 'steering' or 'acceleration/deceleration'. The driver must handle another task simultaneously. Driving Role: The driver remains the primary operator, needing to stay focused and ready to take over at any time. Common examples: . Adaptive Cruise Control (ACC): The system can automatically accelerate and decelerate to maintain distance from the car in front. . Lane Keeping Assistance (LKA): The system can assist in steering to keep the vehicle within the lane. Level 2: Partial Automation System Capability: The system can assist with both 'steering' and 'acceleration/deceleration' tasks simultaneously. This is currently the most mainstream level of driving assistance in the market. Driving Role: The driver must monitor the driving environment at all times and be ready to take immediate control. Although the system appears to drive itself, the driver must keep their hands on the steering wheel and focus on the road. Legal and system responsibility still lies with the driver. Common examples: . Tesla's FSD . Ford's Co-Pilot360 . Volvo's Pilot Assist Key distinction: The responsibility from Level 0 to Level 2 lies entirely with the human driver. Even when the system is operational, the driver must monitor continuously and assume full responsibility. Level 3: Conditional Automation System Capability: Under specific and limited conditions (e.g., highways, good weather, low-speed congested areas), the system can fully perform all driving tasks, including monitoring the surrounding environment. Driving Role: Under the conditions when the system is activated, the driver can temporarily divert their attention from driving tasks (e.g., checking their phone or watching a video). However, the driver must remain alert and be able to regain control of the vehicle immediately when the system requests takeover. Responsibility Transfer: This is the first level where responsibility begins to shift. During system operation, the vehicle assumes driving responsibility; but when the system requests takeover, responsibility returns to the human driver. Common examples: . Mercedes-Benz's DRIVE PILOT (legally operational in parts of Germany and some US states) . Honda SENSING Elite (equipped in the Legend model for the Japanese market) Level 4: High Automation System Capability: Under most conditions, the system can fully drive autonomously and can handle system failures or abnormal situations on its own. Driving Role: Within the design operational range, the driver does not need to intervene or monitor at all. The vehicle can even be designed without a steering wheel or pedals. If the vehicle leaves its designed operational range (e.g., enters a rural road or harsh weather), the system will safely pull the vehicle over without requiring human takeover. Responsibility Transfer: Within the design operational range, responsibility is fully assumed by the system. Common examples: . Tesla's unmanned taxi (Robotaxi) service . Baidu's Apollo unmanned bus . Automated valet parking systems (AVP) in limited areas Level 5: Full Automation System Capability: At any time, in any location, and under any weather or road conditions, the system can perform all driving tasks. It can do everything that an experienced human driver can do. Driving Role: No human driver is required in the vehicle, and there may not even be any passengers. The design of the vehicle will no longer require a steering wheel, accelerator, or brake pedal. Responsibility Transfer: Responsibility always lies with the autonomous driving system. Common examples: There are currently no commercialized or realized Level 5 vehicles; this remains the ultimate goal of autonomous driving technology. Current usage status and adoption rates of FSD in various countries As of the end of August 2025, the usage of FSD shows rapid progress, but adoption rates still face challenges. Tesla reports indicate that FSD (Supervised) has accumulated approximately 4.74 billion miles of driving mileage, demonstrating widespread practical application. In terms of safety data, the vehicle safety report for the second quarter of 2025 indicated that vehicles using FSD suffer one collision for every 10 million miles driven, while the average for vehicles not using the assistance system is one for every 1.3 million miles, making FSD about 10 times safer than average driving safety in the US. This is due to FSD not experiencing fatigue or driving under the influence and its ability to handle complex situations. However, adoption rates remain low. A survey this year showed that only 14% of American consumers indicated that FSD would make them more likely to purchase a Tesla, while 36% of...

#AI