The Golden Age of HD Mapping for Autonomous Driving
The market for Advanced Driver Assistance Systems (ADAS) and Autonomous Vehicles (AV) will rise from US$3 billion in 2015 to US$96 billion by 2025, according to a Goldman Sachs report. As an integral part of the system, High Definition (HD) Maps bring functions such as high-precision localization, environment perception, planning and decision making, and real-time navigation cloud services to autonomous vehicles.
Despite the laborious and expensive process involved, startups, big tech, and even auto-parts manufacturers are increasingly investing in the production of HD Maps. This spring Baidu’s million-dollar mapping vehicles began accurate and precise localization and data collection in a 20-km (12-mile) radius of a Beijing park for centimeter-specific HD Maps. The team spent five days on fixed GPS, and a half day driving the area 5 to 10 times with a 64-channel LiDAR system.
Competition is especially intense among startups, where funding and finding an effective strategy are critical for early-stage business growth. HD Mapping startups have turned to crowdsourcing, sought to lower the cost of HD mapping software and hardware, or focused only on technical services for HD Map development.
Below we look at the pros and cons of different approaches for entering the HD Mapping market, along with sample cases.
Strategy A: Mapping Service Provider
A rising US star in HD Mapping is DeepMap. The company provides hardware tools, software solutions, field data collection services, and can transform customers’ self-driving fleet data into their own personalized HD maps. An insider told Synced that DeepMap charges about US$5,000 per kilometer for mapping services in the US. The company has recently expanded its business to China.
- Fast product and service commercialization for business growth.
- Requires relatively low surveying and mapping capabilities.
- Technological plans or mapping services are only part of a complete HD Mapping solution. Such services are also not universally applicable. Chinese map provider Kuandeng’s CTO Hanping Feng told Synced, “Most domestic auto manufacturers in China prefer to find one partner who can provide HD maps in mass production, rather than distribute assignments to several different companies.”
Comment: Providing tech or solutions can be a shortcut for startups eager to enter the HD Mapping market. However, such companies will only partially meet market and customer requirements.
Strategy B: Crowdsourcing
IvI5 is a US startup that applies computer vision technology for HD Mapping. For data gathering, the company encourages Uber drivers to use its app Payver, “the dashcam that pays you to drive.” Drivers who record videos while working accumulate points they can trade for rewards.
Mapbox, another US startup, gets its data sources through the open map community OpenStreetMap. The company provides developers with HD map products, and developers return the favour by uploading data to the platform.
- The least expensive strategy for HD mapping so far.
- An effective and efficient method for HD map updates.
- High threshold: Map crowdsourcing requires up to millions of participants for maximum value generation.
- Data (such as road signs and pavement markings) collected with cameras and supporting software may be fragmented or redundant.
- Crowdsourced data from open-source platforms has varied quality, and so requires data filtering and post-processing.
Comment: Even though crowdsourcing is cost-effective, human experts and high precision surveying and mapping equipment are still essential for HD Map creation. No current Chinese HD Map startups are pursuing crowdsourced mapping solutions.
Strategy C: Hardware and Software Cost Reduction
Chinese map provider Kuandeng offers a full-stack solution for HD map production with a strategy similar to AutoNavi and Baidu. However while these tech giants use expensive LiDAR equipment for field mapping, Kuandeng has found a way to reduce hardware and software costs by an order of magnitude.
“We tried a variety of hardware and software solutions for data acquisition and have proved that computer vision technology has a comparable precision with LiDAR in highway scenes,” says company CTO Hanping Feng. “Moreover, computer vision is more helpful than LiDAR for localization, it can for example accurately record the starting point of each dashed line pavement marking using a visual scheme.”
- A complete map product is more attractive to the market and customers.
- This strategy avoids the crowdsourcing problems, as professional surveying and mapping personnel and equipment can guarantee accuracy, precision and integrity of map information.
- The cost is greatly reduced compared with traditional LiDAR techniques.
- Difficult for a startup to directly compete with established mapping companies.
- Not suitable for timely HD Map updating, where crowdsourcing is a better solution.
Comment: Startups focusing on this strategy have their own HD Map products, which can help secure partnerships with automakers to quickly scale the companies and increase market share.
Looking back at the global development of traditional digital maps and navigation, we see that dozens of companies appeared early and experienced rapid growth — but in the end just a handful came to share over 95 percent of the market. There are many factors to consider in determining the best strategy for tapping into the rapidly evolving HD Mapping market. But if history repeats, a company that scales quickly should have a better chance at long-term success.
Source: Synced China
Localization: Tingting Cao | Editor: Michael Sarazen
Follow us on Twitter @Synced_Global for more AI updates!