You May Also Like
Tesla’s Full Self-Driving (FSD) driver-assistance software, FSD V14, has improved its ability to respond to unexpected situations such as roadworks, pedestrians and obstacles on the road, an evaluation found. The review also said it still needs improvement in reaching destinations and parking, and controversy has emerged over collision-avoidance features being tied to a paid subscription.
Electrek media outlet CleanTechnica reported on Oct. 28 (local time) that a Tesla user driving a 2026 base Model Y recently shared real-world experience with FSD V14 and assessed that situations requiring driver intervention fell sharply from departure to arrival.
The most noticeable change was its handling of so-called “edge cases.” Edge cases refer to exceptional situations that are hard to predict, unlike typical driving conditions. The user explained that FSD found a route and drove even when construction narrowed lanes or required detours into the opposite lane. It also stopped when pedestrians showed signs of crossing at crosswalks, and slowed or avoided deer by the roadside or obstacles in a lane, the user said.
A newly highlighted function was “automatic collision avoidance.” Under this approach, the system automatically intervenes and performs an evasive manoeuvre if a collision risk is imminent even if the driver did not activate FSD during that drive. The user said the feature requires a $99-a-month FSD subscription and activation in a related menu.
Improvements were also observed in driving convenience. FSD recognised flashing school-zone signals and slowed to 20 miles per hour, and it reduced speed not only for speed bumps but also for deeply sunken drainage sections of road, the user said. The user said the car-summon function worked reliably in a large parking lot, and that on a four-lane highway it changed lanes in advance to overtake and then returned to the right lane after completing the manoeuvre. The user added that they had not experienced “phantom braking,” in which the vehicle brakes suddenly despite no obvious obstacle, for a long time.
But the improvements were not complete in all driving situations. The biggest tasks were cited as driving just before reaching a destination and parking. The review said more improvement is needed for functions such as entering a garage on its own, choosing a parking spot near a desired entrance at a large store and exiting complex parking lots reliably.
Limits were also cited in detailed driving actions. The user said the system came to a complete stop at stop signs and then waited about 1.5 seconds, and that at signalised intersections it sometimes entered a turning lane by mistake when it should have gone straight.
The paid nature of safety functions is also a point of debate. The user assessed FSD as a safety feature that can save lives, but criticised that drivers who do not pay a monthly subscription cannot use automatic collision avoidance. The user argued there needs to be an option to provide only safety-critical functions such as collision avoidance separately, even without subscribing to the full driver-assistance package.
The user also suggested changing the setup so FSD use is enabled by default and can be turned off when needed, rather than requiring drivers to switch it on each time. The user said the need for driver intervention fell sharply on highways and also suggested it could be possible to review approval for a higher level of automation, even in limited sections. The user added that this was a personal evaluation and proposal and does not mean FSD has been approved as a self-driving system that can operate without driver supervision.
Another point of interest is the share of Tesla buyers choosing FSD. Figures presented in the evaluation showed the share of new-car buyers selecting FSD rose to as high as 75 percent in the third quarter of 2026, a sharp increase from about 12 percent in the past. But the basis and scope of the calculation need separate verification.
The evaluation is based on a single user’s real-world driving experience, not an official performance verification. As a result, it is difficult to generalise improvements seen in a specific vehicle and driving environment as reflecting all users’ experience or an overall increase in safety. Even so, its response in exceptional situations such as roadworks, pedestrians and obstacles is drawing attention as a key indicator of FSD’s real-world competitiveness.
The key going forward is whether it can improve completion in the final stages of driving, including parking and reaching destinations, while also defining how safety functions are provided and how responsibility for driver supervision is established. In particular, the paid subscription policy for collision avoidance is expected to lead to debate not only over technical performance but also access to safety functions as FSD adoption expands.
