An advanced driver-assistance system (ADAS) has turned out to be a must-have for car buyers, as eight out of 10 are willing to pay for the feature, said Zhang Kai, the chairman of Beijing-based autonomous driving company Haomo.AI Technology Co., Ltd., at the firm’s seventh AI Day event held on January 5.
Zhang also shared with participants his insights on the top 10 trends in the autonomous driving industry.
Zhang Kai, the chairman of Haomo.AI, speaks at the company’s AI Day event on January 5, 2023.
Autonomous driving sector will fully enter the 3.0 era
Tesla has moved into the data-driven autonomous driving 3.0 era, away from the hardware-driven 1.0 era and the software-driven 2.0 era, while companies in China that have chosen a “progressive” path to full automation, represented by Haomo, will enter this stage in 2023 when they achieve the data closed loop via mass production and application of ADAS.
Intelligent driving will become a standard feature for mid-range passenger vehicles
An ADAS feature that relies on ordinary navigation maps, rather than high-definition maps, can reduce costs so much that it will be possible for automakers to provide that as a standard feature in medium-level vehicles.
Great Wall Motor has equipped Haomo’s ADAS feature in more than 20 models so far.
New-generation chips and sensors will enter the market
With the rapid surge in the penetration rate of intelligent cars in China, chip and sensor shipments are exhibiting strong growth.
In 2023, a new generation of AI chips in China and abroad, as well as 4D millimeter-wave radar and blind-compensating LiDar (Light Detection and Ranging) will enter the market in a major way, providing more choices for autonomous driving companies.
Big AI models will speed up autonomous driving evolution
Transformer, which has been applied by Haomo as well as more and more other companies to train their algorithms, can cut costs in data synthesis and knowledge extraction and speed up the upgrade of autonomous driving technologies tremendously.
For users, efficiency in commuting matters more than functions
Driven by real user data, the iteration of intelligent driving systems is more accurate, and the iteration speed is also increasing fast. As a result, the race to achieve highly intelligent driving abilities will change from a competition for functional completeness to a competition for improving commuting efficiency.
Supercomputing centers will be an entry-level requirement for autonomous driving companies
As autonomous driving enterprises move toward the 3.0 era, the data-driven mode of “big model + big data” becomes the key to the evolution of autonomous driving technology. As a result, the supercomputing center that drives the training of big models and massive amounts of data will become the entry-level requirement for autonomous driving enterprises.
Haomo launched MANA OASIS, China’s largest self-driving computing center, at the AI Day event, along with Volcano Engine, a cloud service platform owned by ByteDance.
Mana Oasis has a total computing power of 670 Pflops, or 670 quadrillion floating-point operations per second, higher than the 600 Pflops of EV maker XPeng.
ADAS solutions for urban roads will be widely available to automakers
In 2022, ADAS for passenger vehicles started to function beyond highways on urban roads on a trial basis. In 2023, such features will work inside city areas on a large scale.
ADAS solutions relying purely on navigation maps, rather than high-definition maps, enable a wide coverage of more cities.
In the first half of 2024, Haomo plans to cover 100 cities in China with its ADAS solution, enabling vehicles to drive and park autonomously in urban areas.
Car owners will rely more on ADAS
As more and more drivers start using ADAS, they will shift from just experiencing innovation in the automotive industry to relying on partial automation.
Great Wall Motor vehicle owners have used HPilot, Haomo’s autonomous driving solution, across a combined distance of 25 million kilometers on Chinese highways.
Cost of an autonomous van will fall below RMB 100,000
Although autonomous driving for the logistics sector has faced some setbacks in the United States, the industry in China has ushered in favorable policy support.
In addition, the cost for a Level 4 autonomous driving van will fall below RMB 100,000.
Currently, Haomo sells such a van, the Xiaomotuo or HDeliver, at RMB 128,888. Moreover, Haomo has also been leveraging Xiaomotuo to provide delivery services to clients. The number of on-demand grocery delivery orders fulfilled by Xiaomotuo to WuMart by the end of 2022 had reached 130,000.
Haomo established cooperation with Nasdaq-listed on-demand services provider Dada Group in November to co-develop unmanned deliveries on a regular basis.
High demand for AI talent will continue
Although technology companies downsized in 2022, autonomous driving companies continued to compete for AI talent.
As automobile intelligentization will remain the main battlefield in the technology sector in 2023 and beyond, the competition for AI autonomous driving talent will persist.
Haomo has started to recruit employees based in Munich, Germany, this year. The company is also seeking talent for R&D positions in Beijing, Baoding, Shenzhen and Shanghai.