As the supervision of HD map qualifications tightens, issues such as map collection cost, update frequency, and coverage stand out. Amid the boom of urban NOA, the 'lightweight map' intelligent driving solution has become a hot topic in 2023. This solution lessens the dependence on offline HD maps, posing a challenge to the development of HD maps.
From the development process of autonomous driving, it can be seen that human-machine co-driving will exist for a period of time. The need for maps in this phase is not necessarily HD maps. Multi-source maps that integrate the complementary characteristics of different maps may be more suitable for the needs of autonomous driving in this phase.
In June 2023, the Map Technology Review Center of the Ministry of Natural Resources announced the phased progress in review of ADAS maps of ordinary urban roads across China, and allowed companies to submit ADAS maps of nationwide ordinary urban roads for review in batches. Currently, NavInfo’s approved nationwide urban ADAS map data have covered 120 cities in 30 provinces; Baidu Maps has ADAS maps of 134 cities approved.
OEMs: relevant departments’ stricter review of the Class A qualification for navigation electronic map surveying and mapping has discouraged OEMs to deploy the Class A qualification for map surveying and mapping. At present, some OEMs use neural network model algorithms for real-time mapping and lower reliance on offline HD maps, and the ADS-enabled models of Tesla, Li Auto, Xpeng, and Huawei are typical cases; some other OEMs prefer stability, and obtain surveying and mapping qualifications by way of applying for Class B qualification or establishing new joint ventures with map providers. For example, GAC together with its partners such as Nanjing Institute of Surveying, Mapping and Geotechnical Surveying Co., Ltd. co-funded 'Guangdong Guangqi Yutu Equity Investment Partnership (Limited Partnership)'; Anhui NIO Smart Mobility Technology Co., Ltd., a subsidiary of NIO, applied for the Class A qualification for Internet map services.
Map providers: to meet the market demand, they launch 'lightweight map' solutions, putting SD data, HD data, LD data, etc. on one map to ensure the continuity of navigation. One example is Tencent which introduced the 'Intelligent Driving Cloud Map' to support the cooperative construction by map providers, automakers, autonomous driving companies and other players, after launching its 'three-in-one' intelligent driving map.
It is mainly emerging carmakers that are more active in 'lightweight map' solutions. One reason is that they implement urban NOA functions very quickly, and HD maps fail to answer their relevant needs.
Xpeng's solution that does not use HD maps has the advantages of 4 to 10 times faster generalization speed, completely solving the problem of data freshness, reducing costs, and popularizing intelligent driving, compared with the solution using HD maps.
The 'no offline HD map' solution implemented by Xpeng relies on XNet to build a 'HD map' in real time.
Li Auto is now promoting the NPN solution, hoping to solve the problem of online map updates.
In terms of OEMs’ solutions, despite less dependence on HD maps, the 'lightweight map' solution has higher requirements for vehicle perception and algorithms.
Conventional map providers launch lightweight autonomous driving map solutions to meet demand.
The voice of OEMs to 'not rely on HD maps' is growing ever louder. To cater to the market demand, conventional map providers also make changes, trying hard to solve the three enduring problems of HD maps: update frequency, coverage area, and cost, and launching map products that more fit in with the current needs of autonomous driving.
Tencent Intelligent Driving Cloud Map features scalable multi-layer forms, covering basic map layer, update element layer, ODD dynamic layer, driving experience layer and operation layer. Automakers can flexibly configure and manage the layers as they need, and build a data-driven operation platform suitable for themselves by combining it with their own data layer.
This product will be delivered within 3-5 business days.
From the development process of autonomous driving, it can be seen that human-machine co-driving will exist for a period of time. The need for maps in this phase is not necessarily HD maps. Multi-source maps that integrate the complementary characteristics of different maps may be more suitable for the needs of autonomous driving in this phase.
How do players respond to the development of new-generation autonomous driving maps?
Government: while tightening the Class A qualification for HD map surveying and mapping, work to enhance the review of ADAS maps and Class B surveying and mapping qualification.In June 2023, the Map Technology Review Center of the Ministry of Natural Resources announced the phased progress in review of ADAS maps of ordinary urban roads across China, and allowed companies to submit ADAS maps of nationwide ordinary urban roads for review in batches. Currently, NavInfo’s approved nationwide urban ADAS map data have covered 120 cities in 30 provinces; Baidu Maps has ADAS maps of 134 cities approved.
OEMs: relevant departments’ stricter review of the Class A qualification for navigation electronic map surveying and mapping has discouraged OEMs to deploy the Class A qualification for map surveying and mapping. At present, some OEMs use neural network model algorithms for real-time mapping and lower reliance on offline HD maps, and the ADS-enabled models of Tesla, Li Auto, Xpeng, and Huawei are typical cases; some other OEMs prefer stability, and obtain surveying and mapping qualifications by way of applying for Class B qualification or establishing new joint ventures with map providers. For example, GAC together with its partners such as Nanjing Institute of Surveying, Mapping and Geotechnical Surveying Co., Ltd. co-funded 'Guangdong Guangqi Yutu Equity Investment Partnership (Limited Partnership)'; Anhui NIO Smart Mobility Technology Co., Ltd., a subsidiary of NIO, applied for the Class A qualification for Internet map services.
Map providers: to meet the market demand, they launch 'lightweight map' solutions, putting SD data, HD data, LD data, etc. on one map to ensure the continuity of navigation. One example is Tencent which introduced the 'Intelligent Driving Cloud Map' to support the cooperative construction by map providers, automakers, autonomous driving companies and other players, after launching its 'three-in-one' intelligent driving map.
Emerging carmakers take the lead in launching 'lightweight map' solutions.
At present, OEMs’ solutions that do not rely on HD maps don’t mean that they do not use maps at all, but subtract elements from HD maps or add them to navigation maps instead.It is mainly emerging carmakers that are more active in 'lightweight map' solutions. One reason is that they implement urban NOA functions very quickly, and HD maps fail to answer their relevant needs.
Xpeng
In the first half of 2023, Xpeng started developing intelligent driving solutions based on SD maps. NGP that uses HD maps or does not use adopts the same technology stack. The only difference is that the original HD map input is replaced by the navigation map input, and the understanding of navigation information in real-time perception.Xpeng's solution that does not use HD maps has the advantages of 4 to 10 times faster generalization speed, completely solving the problem of data freshness, reducing costs, and popularizing intelligent driving, compared with the solution using HD maps.
The 'no offline HD map' solution implemented by Xpeng relies on XNet to build a 'HD map' in real time.
Li Auto
Li Auto has launched urban NOA in 2023. This solution does not rely on HD maps. It aims to construct the features of intersections to assist in real-time perception and mapping. In a word, road sections are 'unmapped', and intersections are mapped by crowdsourcing.Li Auto is now promoting the NPN solution, hoping to solve the problem of online map updates.
In terms of OEMs’ solutions, despite less dependence on HD maps, the 'lightweight map' solution has higher requirements for vehicle perception and algorithms.
Conventional map providers launch lightweight autonomous driving map solutions to meet demand.
The voice of OEMs to 'not rely on HD maps' is growing ever louder. To cater to the market demand, conventional map providers also make changes, trying hard to solve the three enduring problems of HD maps: update frequency, coverage area, and cost, and launching map products that more fit in with the current needs of autonomous driving.
Baidu
In July 2023, Baidu MapAuto 6.5, a human-machine co-driving map, was launched. It is a full 3D lane-level map and also an all-scenario human-machine co-driving map. It can provide three types of data: SD, LD and HD. Wherein, SD data has covered the whole country and is currently available on 10 million vehicles. Baidu’s LD lightweight map data service consists of lane-level topology, complex scene geometry, experience layer, and dynamic information layer, allowing for daily update.Amap
The new HQ Live MAP, launched in June 2023, combines the merits of HD MAP and SD MAP. In spite of a lower accuracy than HD MAP (absolute accuracy: 50cm, relative accuracy: 10cm), HQ Live MAP is enough for ADAS scenarios (highway and urban expressway scenarios: absolute accuracy of 1m, and relative accuracy of 30cm; ordinary urban road scenarios: relative accuracy of 1m), and it also simplifies unnecessary map elements in ordinary urban road scenarios, further reducing production and deployment costs.Tencent
The latest Intelligent Driving Cloud Map, released in September 2023, enables fully cloud-based autonomous driving maps, supports element-level and minute-level online updates, and allows for the cooperative construction by map providers, automakers, autonomous driving companies and other players.Tencent Intelligent Driving Cloud Map features scalable multi-layer forms, covering basic map layer, update element layer, ODD dynamic layer, driving experience layer and operation layer. Automakers can flexibly configure and manage the layers as they need, and build a data-driven operation platform suitable for themselves by combining it with their own data layer.
Autonomous Driving Map Industry Report，2024 highlights the following:
- Autonomous driving map (formulation of policies, regulations, standards, etc.);
- Vehicle map amid the development of urban NOA (development direction, coping strategies of conventional map providers, main types of maps used in urban NOA, etc.);
- HD map (market status, market size, company pattern, business model, development challenges, etc.);
- Application scenarios of intelligent driving map (high-speed autonomous driving of passenger cars, low-speed parking, autonomous human carrying, autonomous object carrying, etc.);
- Major Chinese and foreign map providers (map product series, new product layout, product application cooperation, etc.);
- HD map technology companies (technology layout, new technology R&D, etc.).
This product will be delivered within 3-5 business days.
Table of Contents
1 Status Quo of Policies, Standards and Regulations Concerning Autonomous Driving Map
2 Status Quo of Autonomous Driving Map Market
3 Status Quo of HD Map Market
4 Intelligent Driving Map Application Layout of OEMs
5 Chinese and Foreign Map Providers
6 HD Map Technology Companies
Companies Mentioned
- Baidu Maps
- NavInfo
- Amap
- Tencent
- BrightMap
- Mxnavi
- Huawei
- Heading Data Intelligence
- JD
- Leador
- eMapgo
- Momenta
- Roadgrids
- Here
- Mobileye
- NVIDIA
- DeepMotion
- Mapbox
Methodology
LOADING...
