Li L9 and Li L8 have brought stronger assisted driving functions, which are also deeply loved by car owners. As the mileage continues to increase, the Li car product team has also received various questions from users about the field of assisted driving.
In this regard, Li Auto selected the five most representative questions to answer, let’s take a look together.
Q: My car has recognized the cone, why can’t the car dodge by itself?
A: There is an order of magnitude difference in the accuracy of recognition between “displaying the identified cone on the screen” and “identifying the cone and controlling the vehicle autonomously”.
At present, we are improving the perception algorithm of this part, and the follow-up (AD Max) will support the ability to avoid the cone scene.
Q: At present, there is still a sense of oppression when passing a large truck on a high speed. How to solve this problem?
A: “Truck avoidance ability” will be significantly improved in OTA 4.2.
Regarding the extent of avoiding trucks, we need to make judgments based on the actual scene, for example, whether there is a car on the other side, whether it is in a curve, OTA 4.2 will combine user feedback, and then improve the subdivision scene to make avoidance more accurate. personification.
Q: High-precision maps are often disconnected on the highway. Why is this?
A: Navigation-assisted driving relies on the coverage of high-precision maps. Li Auto uses AutoNavi high-precision maps with the highest product power in the industry. However, because high-precision maps are in mass production, they are still in an industry compared to navigation maps. The initial state (Legal ONE 2021 and L9/L8/L7 are equipped with navigation-assisted driving as standard, and Li Auto is currently the car company with the highest number of users of domestic navigation-assisted driving, but from the perspective of the overall automobile industry, those with this capability models are still very few).
At this stage, the update frequency of high-precision maps is not comparable to the frequency of road construction and rerouting of actual Chinese roads. Once the data of the high-precision map differs from the real world, or the confidence level decreases, we will downgrade the navigation-assisted driving to the lane-keeping assist function to ensure the safety of the scene.
In the future, we will optimize the “disconnected” experience by strengthening the real-time processing capabilities of visual perception and cooperating with AutoNavi to improve the accuracy of high-precision maps.
Q: Can I still optimize the long start time for a straight line calls?
A: The long start-up time of AD Max is mainly due to the long start-up time of Orin-X when the vehicle is cold-started, which takes about 40 seconds. At present, we have an optimization strategy, which is under development and will be launched in Q1 next year to reduce the start-up time of straight-line summoning to within 10 seconds.
Q： What should I do if the automatic lane change overtaking with NOA is ineffective and the efficiency of passing is reduced instead?
A: In fact, NOA’s [lane change aggressiveness] and [invalid lane change] have been two dimensions that we have to keep in balance during the R&D process. In the premise that assisted driving does not yet have the perception and prediction ability of human drivers, these two performances are basically confronting each other.
Many times human drivers rely on experience and intuition to avoid ineffective lane changes when their vision is obscured.
Later we will further improve the perceived distance and the prediction ability to surround vehicles to achieve a better balance between [lane change aggressiveness] and [invalid lane change], but cannot completely avoid [invalid lane change]. In fact, if you observe the overtaking behavior of human drivers in high speed, there will be [invalid lane change], but people will quickly correct his decision.
We will let the algorithm learn from people for a long time in the future, and iterate to be closer to the human decision making ability.
Li Auto currently has launched three models: Li L9, Li L8, and Li L7.
Li L9 launched a total of 1 model, positioned as the flagship six-seat large SUV, priced at 459,800 yuan; Li L8 is the successor of Li One, a total of 2 models, priced at 359,800 yuan/399,800 yuan respectively, positioned as a luxury six-seat medium and large SUV; Li L7 also includes 2 models, priced at RMB 339,800/RMB 379,800, which are positioned as the flagship five-seater medium-to-large SUV; and the next Li L6 is positioned as a luxury five-seater medium-sized SUV, which is expected to be within RMB 300,000.
