December 3, 2022

Analysis on Rising Automaker Technique: the strategic format of Li Auto in electrical cars, cockpits and self sustaining riding

Li Auto will shift from the only extended-range path to the “extended-range + high-voltage battery-electric” course of in 2023.

Within the first part of 2022, Li Auto offered 60,801 cars, up 99.1% year-on-year. When it comes to fashions, Li ONE nonetheless performed a major function within the first part of the yr. With the release of L8 in September 2022, Li ONE might be regularly withdrawn from the manufacturing line, whilst L8, L7 and L9 might be the point of interest of manufacturing and advertising and marketing someday.

Li Auto gross sales Jan-June 2022

As for product making plans, all of the fashions these days being offered by means of Li Auto are extended-range electrical cars. Then again, Li Auto plans to release no less than two high-voltage battery-electric cars once a year from 2023 onward. For the aim of high-voltage super-fast charging, Li Auto deploys the next 4 sides: First, 4C batteries. 2nd, utility of SiC generation. 3rd, thermal control machine. Fourth, 400KW charging community.

Consistent with its plan, Li Auto will produce the third-generation semiconductor SiC energy chip in 2024. On the identical present of the high-voltage platform, this chip is 70% smaller than an IGBT chip, with the great potency being stepped forward by means of 6%. The format of Li Auto’s 800V high-voltage battery-electric generation unearths that some of the promoting issues of recent vehicles someday might be mirrored within the charging pace.

Li Auto has self-developed AEB and NOA and laid out self sustaining riding chips to growth on clever riding
 
As for the growth of clever riding, Li Auto has evolved AEB machine on its own as a “latecomer”. Sooner or later, Li Auto will supply all open supply codes of its AEB machine to beef up site visitors protection.
 
As well as, Li Auto added NOA to 2021 Li ONE in December 2021, stepped forward the efficiency of AEB, and optimize the detection and fusion of cameras and radar. Since 2022, all new vehicles had been supplied with Li Auto’s self-developed {hardware} suitable with L4 self sustaining riding as usual. Li Auto plans to make city NGP purposes to be had in Li L9 via OTA in 2023, and set up L4 self sustaining riding capacity on manufacturing cars by means of OTA round 2024.

See also  Nvidia's GPU export restrictions might obstruct self-driving R&D for Chinese language EV makers

In regards to the core underlying generation format of clever riding, Li Auto established Sichuan Lixiang Clever Era Co., Ltd. in Might 2022 to design chips. In August 2022, Xie Yan, the previous vp of Huawei Tool, joined Li Auto as the top of machine R&D department. The machine R&D department is principally accountable for R&D of a few underlying clever applied sciences, together with Li Auto’s self-developed running machine and computing platform. Li Auto’s computing platform industry additionally comprises its self-developed clever riding chip mission.
 
For the clever riding set of rules, Li Auto makes use of BEV framework very similar to that of Tesla, this is, it makes use of natural imaginative and prescient for movement belief prediction. At the foundation of BEV visible data, Li Auto exploits further LiDAR and HD map data enter to put into effect the BEV fusion set of rules, and provides a visible safety module and a LiDAR safety module that are redundant with BEV framework style for the sake of an additional layer of coverage.

The cockpit of Li Auto upgrades from 2D interplay to three-D interplay.

The cockpit multi-modal interplay represents building development of human-machine co-driving generation. As in keeping with 3 new vehicles introduced in 2022, Li Auto upgrades the previous four-screen 2D interplay in Li ONE to present five-screen three-D interplay, and realizes “voice+gesture” multi-modal interplay.

As an example, the 5 displays of Li L9 come with a protected riding interactive display, a W-HUD with a projected house of 13.35 inches, a fifteen.7-inch built-in heart console display and co-driver display, and a fifteen.7-inch rear leisure display. The in-vehicle three-D ToF sensor perceives the cockpit surroundings in actual time. Plus, 6 microphones, 7.3.4 panoramic sound format, 5G dual-operator automobile communique community, and multi-modal spatial interplay generation evolved by means of Li Auto in keeping with deep finding out allow the three-d interplay within the cockpit.

See also  China state paper article explains why Audi's NEVs do not outsell NIO, Li Auto

When it comes to belief, Li AI, the clever cockpit house, imitates the coordination of human ears and eyes to score the three-d data belief throughout the automobile underneath the affect of multi-modal consideration generation by means of a disbursed hexasilicon microphone, an IR three-D ToF sensor, MIMO-Internet six-vocal-range enhancement community and MVS-Internet multinocular & multi-view visible fusion community.
 
As for working out and expression, Li AI restores the multi-source heterogeneous information sensed by means of fusion belief to concrete occasions within the community, and fulfills additional summary working out. In the end, wisdom linking, wisdom of entirety and logical reasoning shape an tournament graph, permitting machines to have their very own working out and decision-making functions.
 
Referring to voice generation, Li Auto defines its voice assistant “Lixiang Tongxue” because the consumer’s housekeeper (present level) and circle of relatives (long run function), and plans a three-stage product improve. At the moment, the objectives of the primary two levels had been accomplished via OTA: The primary level: Li Auto’s self-developed “Lixiang Tongxue” engine replaces the underlying functions with Horizon + AIspeech + Microsoft, and so forth.

The second one level: “what you spot is what you’ll be able to say”, four-vocal-range locking and different purposes.
 
Sooner or later, the voice machine will be offering purposes reminiscent of “from application-on-demand to network-on-demand”, cross-screen multi-person discussion, and “the entrance passenger can pick out up the dialog after the motive force finishes talking”.