[ad_1]
- Baidu plans to ascertain the world’s largest absolutely driverless ride-hailing service space in 2023
- Baidu continues to point out progress momentum because the world’s largest robotaxi service supplier, reaching 1.4 million orders by Q3 2022
- Baidu unveiled a collection of autonomous driving know-how breakthroughs together with the {industry}’s first massive mannequin constructed for autonomous driving, aiming to eradicate longstanding industry-wide challenges in automobile recognition
BEIJING, Nov. 29, 2022 /PRNewswire/ — Baidu, Inc. (NASDAQ: BIDU and HKEX: 9888) (“Baidu” or the “Firm”), a number one AI firm with sturdy web basis, right now introduced plans to construct the world’s largest autonomous ride-hailing service space in 2023. The announcement got here throughout a celebration of Apollo Day, Baidu’s autonomous driving tech occasion (Click to replay 2022 Baidu Apollo Day). The plan outlined a purpose to broaden the operation space for Baidu’s absolutely driverless robotaxis, permitting Baidu to succeed in extra potential clients. Moreover, Baidu revealed a collection of latest know-how breakthroughs together with an AI massive mannequin constructed for autonomous driving notion, high-definition autonomous driving maps, a closed-loop autonomous driving knowledge system, and the profitable end-to-end adaptation of AI chips for autonomous autos.
Baidu to construct the world’s largest absolutely driverless ride-hailing service space in 2023, sustaining progress momentum because the world’s largest robotaxi service supplier
Since August 2022, Baidu has already rolled out absolutely driverless ride-hailing companies (with no human drivers within the automotive) within the cities of Chongqing and Wuhan, with entry to lots of of sq. kilometers of operation space. Based mostly on this main place, Baidu will proceed to broaden its operation space subsequent 12 months to construct the world’s largest service space for absolutely driverless robotaxi service.
At present, Baidu’s autonomous ride-hailing platform Apollo Go covers greater than 10 cities in China together with all first-tier cities. In Q3 2022 alone, Apollo Go has accomplished greater than 474,000 rides, up 311% 12 months over 12 months, and a 65% improve in comparison with final quarter. In first-tier cities like Beijing and Shanghai, every robotaxi on Apollo Go can present 15 rides a day on common, almost the identical day by day trip common of typical on-line ride-hailing automotive companies. By the top of Q3 2022, the accrued rides offered to the general public by Apollo Go have reached 1.4 million. As Baidu continues to scale up its operation space of robotaxi service, it’s one step nearer to the purpose of offering autonomous driving companies to extra individuals, whereas additional strengthening its main place within the international autonomous ride-hailing market.
“Backed by its strong AI know-how, Baidu Apollo has created a protected, clever and environment friendly autonomous driving know-how system, bringing robotaxi companies from designated zones to open roads at scale,” stated Jingkai Chen, Baidu’s autonomous driving know-how skilled on the occasion. The generalization capacity of Baidu’s autonomous driving know-how has progressed at a extra superior tempo than anticipated. Now, the lead time to deploy autonomous driving know-how in a brand new metropolis is simply 20 days.
Trade’s first AI massive mannequin for autonomous driving, addressing the “lengthy tail” downside
The AV {industry} has lengthy grappled with the “lengthy tail” downside, through which an autonomous automobile runs right into a situation it has not seen or skilled earlier than. To deal with this downside, Baidu’s autonomous driving know-how skilled Jingdong Wang has introduced the {industry}’s first AI massive mannequin for autonomous driving, a pre-trained visual-language mannequin with weak supervision, backed by the Baidu WenXin Large Mannequin, which acknowledges hundreds of objects, serving to to enlarge the scope of semantic recognition. The mannequin will allow autonomous autos to shortly make sense of an unseen object, reminiscent of particular automobile (fireplace truck, ambulance) recognition, plastic bag misdetection, and others. As well as, Baidu’s autonomous driving notion mannequin—a sub-model of the WenXin Large Mannequin—leveraging greater than 1 billion parameters, is ready to dramatically enhance the generalization potential of autonomous driving notion.
Excessive-definition autonomous driving map to safeguard a wiser and extra environment friendly autonomous driving expertise
“Baidu’s new era autonomous driving map is provided with complete capabilities reminiscent of automated manufacturing, real-time fusion, and information enhancement.” in response to Jizhou Huang, Baidu’s autonomous driving know-how skilled. Will probably be put into mass manufacturing to understand a “protected, dependable and environment friendly” autonomous driving expertise.
- 96% automation fee for map manufacturing: With AI as the important thing driving power to extend effectivity and produce down the associated fee, the automation fee of Baidu’s excessive precision map manufacturing has now reached 96%.
- Actual-time map replace to make sure driving security: The combination of vehicle-side notion knowledge and multi-source maps to generate on-line maps in actual time has helped to considerably guarantee the security of autonomous driving.
- Large knowledge plus human driver information to enhance driving reliability: With greater than 12 million kilometers of street networks and knowledge accrued on Baidu Maps, together with lots of of thousands and thousands of human drive-hours, Baidu’s autonomous driving map successfully integrates these knowledge factors to enhance the reliability of autonomous driving.
Closed-loop knowledge system to additional improve the intelligence of autonomous driving
With autonomous autos being deployed at a bigger scale, the dimensions of knowledge acquired may even improve exponentially. This brings challenges to figuring out helpful knowledge and effectively utilizing knowledge to constantly enhance autonomous driving. Ang Li, Baidu’s know-how skilled, launched the idea of “High-quality Purification, Sturdy Ingestion” and Apollo Loop, a closed-loop knowledge system, to successfully establish and make the most of knowledge. To purify the information, the system leverages each on-board small AI fashions and cloud-based massive AI mannequin to realize high-efficiency knowledge mining and automatic labeling. The information ingestion structure achieves automated coaching with its group-optimization capacity and knowledge distribution understanding to successfully make the most of knowledge and additional improve the general intelligence of autonomous driving.
Creating a mutually reinforcing parallel use mannequin for L4 and L2+ autonomous driving applied sciences
Baidu can be actively bringing autonomous driving know-how to empower superior assisted driving merchandise. Baidu’s autonomous driving know-how skilled Liang Wang defined how Baidu is leveraging its decade-long expertise in autonomous driving to discover a technical route by way of which L4 and L2+ autonomous driving can coexist. At present, the know-how stack stage permits the unification of L4 and L2+ good driving merchandise when it comes to visible notion scheme, technical structure, map unification, knowledge interconnection and infrastructure sharing. Baidu envisions a mutually helpful relationship through which L4 will proceed to supply superior know-how migration for L2+ good driving merchandise in city use circumstances, whereas L2 knowledge suggestions may even assist to enhance L4 generalization capacity.
Baidu’s Kunlun AI chip accomplished end-to-end autonomous driving real-world testing
Jian Ouyang, CEO of Kunlun Chip, additionally revealed at right now’s occasion that Baidu’s 2nd-gen Kunlun AI chip has accomplished an end-to-end efficiency adaptation for autonomous driving. This marks a significant milestone additional solidifying and integrating Baidu’s benefits in each autonomous driving software program and {hardware}.
About Baidu
Based in 2000, Baidu’s mission is to make the difficult world easier by way of know-how. Baidu is a number one AI firm with sturdy Web basis, buying and selling on the NASDAQ underneath “BIDU” and HKEX underneath “9888.” One Baidu ADS represents eight Class A abnormal shares.
Media Contact
[email protected]
SOURCE Baidu, Inc.
[ad_2]
Source link