Foreign AV Tech Companies Can Leverage Baidu’s Apollo Platform to Enter China

Figure 2. Apollo Platform Architecture (Version 3.0)

What is Apollo?

Since first announcing its “Apollo” platform for the autonomous vehicle (AV) industry last year, Baidu has consistently upgraded the platform and added new partners and capabilities. As Apollo continues to improve over the coming several years, nTheta anticipates opportunities for foreign AV technology companies to leverage Apollo as a fast track to China’s growing AV market.

On April 18, 2017, Lu Qi (Baidu’s former COO who also heads the Apollo Project) introduced Apollo as “Android for Automatic Vehicles (AV)”. The goal of the Apollo Project is to build “a complete open automatic driving ecosystem” to help manufactures in the automotive industry and autonomous driving field to combine vehicle software and hardware systems to quickly build their own complete AV system.

Figure 1. Former Baidu COO-LuQi
Figure 1. Former Baidu COO-LuQi

The Apollo platform has now experienced four major upgrades ­ from version 1.0 to version 2.5. New components (including hardware sensors and software algorithms) have been added to Apollo, and the latest version (3.0) upgrades the whole platform architecture in an effort to provide a complete development platform with “Mass­-Production” capability.

Apollo Platform's Architecture and Reliability

The growing capabilities of the Apollo platform can be seen in the recently released upgraded platform architecture for Apollo 3.0 See Figure 2.

Figure 2. Apollo Platform Architecture (Version 3.0)
Figure 2. Apollo Platform Architecture (Version 3.0)
  • “Turnkey” Solution First, Baidu promotes Apollo 3.0 as a “TurnKey Solution.” In fact, this solution is actually Apollo’s ultimate goal. But the goal appears well within Apollo’s reach. Baidu has already released into mass production its own automatic vehicle with “L4 Autonomy”­ Apolong . Although mass production by Baidu is not necessarily the same as a turn-key solution for other companies, Apolong does demonstrates that Apollo can be a useful platform for AV production.
  • Cloud Service Platform Apollo’s Cloud Service provides many important modules for AV production, such as HD Map, a Simulation Platform, Data Platform, and Baidu’s DuerOS AI application. These modules are essential for mass production. Furthermore, the Cloud Service makes it easier for Apollo to adopt additional components to either speed up the production process or stabilize the system itself.
  • Open Software Platform “Open Source” is the hallmark of the Apollo platform. It not only can attract AV providers to join the development process, but also can make the Apollo become more “versatile”. Developers can share their automatic driving issues and solutions to help the platform grow quickly.
  • Hardware Development Platform Apollo also provides a Hardware Development Platform, including a hardware reference design, with vehicle selection, key hardware components, peripherals, and a hardware installation guide.
  • Open Vehicle Certificate Platform The open vehicle certification platform proposes a standardized interface between the Autonomous Driving system and the vehicle. Through this platform, AV providers can connect their vehicle to the Apollo open platform, link-up with other AV developers, and accelerate their development. Furthermore, the certification platform provides AV developers with more vehicle options. Using a standardized interface, developers can deploy the same code to multiple Apollo­compatible open vehicles.
While the Apollo 3.0 Architecture appears quite impressive, the key question is whether this type of open source architecture is reliable for autonomous driving? For Apollo, reliability should be a strong point. Automatic driving technology relies on accurate mapping functionality, which is Baidu’s core strength. Baidu Map is one of the top mapping and navigation applications in China. It has achieved more than 1 million valid downloads in Google Play Store. In fact, most Automatic Driving technologies rely on High-Precision Map service, and Apollo 2.0 is actually built upon Baidu Map (See Figure 3).
Figure 3. Apollo 2.0 Software Architecture
Apollo 3.0 uses a relative map module to reduce the dependency on HD Map and make Apollo more accessible to developers. However, relative maps can only be used when the road condition is relatively simple. Thus, HD Map still plays an important role within the whole system. From this point of view, Baidu’s deep experience in maps and navigation services provides the company with a solid foundation for the Apollo Platform.

Behind Apollo

Partnerships and “Big” Driver Data
Currently, Apollo has already attracted over 100 global partners. This partner network covers a broad range of automobile vertical industries, including parts suppliers, OEMs, chip manufacturers, and even some technology giants. See Figure 4.
Figure 4. Part of Apollo's Partners
Figure 4. Part of Apollo's Partners

This wide network of partnerships can help Apollo get what it lacks: Driver Data. Automatic driving requires abundant data for simulation and testing. Even though Baidu excels at HD mapping and AI, without sufficient driver data, Apollo could never compete with competitors such as Waymo and Tesla, which have already collected sufficient driver data from vehicles in various road conditions.

In order to obtain the needed driver data from its partners, Baidu launched an Open Source data set within the Apollo platform named “ApolloScape.” Compared to other open source datasets, such as Kitti and CityScape, ApolloScape is 10x larger in volume, and it is deployed with semantic segmentation to deeply analyze driver video (the most important part of Driver Data). Apollo’s Partners can use this kind of service freely as long as they input their own datasets into ApolloScape.

Platform Monetization
Does open source mean profitless? In fact, Robin Li (CEO of Baidu) discussed how the Apollo platform can become profitable in his statements following the release of Baidu’s Q2 financial report.
First of all, both Apollo and DuerOS are currently in the early stage, and will contribute nothing to profit in a long period of time. However, it can be rather easy to make it profitable when they turn into mature platforms. Baidu can sell HD mapping service, simulation platform, even hardware ACU (Arithmetic Calculation Unit) which are essential for automatic driving”.

– Robin Li – CEO of Baidu

As is discussed above, HD maps are critical for directing self-driving vehicles since it contains accurate information about lane sizes, intersections, road size etc. Baidu has been working on HD maps since 2013. What’s more, Baidu partnered with navigation device maker TomTom to integrate HD maps into the Apollo Platform. This partnership will help Apollo to provide HD maps for US and Western Europe area.
Figure 5. HD Map - Core Part of Automatic Driving
Figure 5. HD Map - Core Part of Automatic Driving

Competition for the AV Market

Although Apollo will certainly face challenges and competition, Baidu has some clear advantages that provide a reasonable basis to believe that Apollo will succeed.

In recent months, the Chinese government has increasingly eased regulations on autonomous vehicles in a number of cities to accelerate the path to commercialization. Now, local governments have the authority to allow autonomous vehicle testing on public roads. Baidu is already testing its software on public roads. In December 2017, Beijing granted Baidu permission to run tests on 33 public roads in the city. On October 29, 2018, Baidu announced that Apollo will also collaborate with Changsha Government to build “The City of AV”.[3]Additionally, when naming the national champions of AI in November, the Chinese Ministry of Science and Technology designated Baidu as the leader in autonomous driving. The other national AI champions include Baidu’s tech peers Alibaba and Tencent — designated as leaders in smart cities and healthcare, respectively[3].

While foreign AV companies have limited access to self-driving tests in China, Tesla just just formed a subsidiary to manufacture its vehicles on mainland China, which would likely increase the number of Tesla vehicles diving on China roads. Given Apollo’s domestic strength, Tesla can definitely extend its China Market by collaborating with Apollo.

nTheta holds the opinion that Apollo provides lots of opportunities to foreign Auto-Vehicle companies, AV sensor (such as LiDAR, Camera etc) manufactures. In fact, Baidu is dedicating significant capital to autonomous vehicle startups, all of which have joined the Apollo partnership alliance.

Apollo Fund was founded in Sept 2017. It is dedicated to financing 100 autonomous driving projects over the next three years. One of its first investments was in Smarter Eye, a software developer focused on ADAS (Advanced Driver Assistance Systems). Baidu led the company’s $15M financing in November. As a matter of fact, Baidu has also invested in startups developing autonomous vehicle hardware. Such as Hesai Tech[4], iDriverPlus[5], Velodyne[6], NIO[7].

Although Baidu was named the leader in autonomous vehicle technology, there are still some notable competitors. Clearly, Alibaba and Tencent are Apollo’s competitors. However, DiDi may become Baidu’s biggest competitor. Although Didi currently provides only ride­hailing service. Didi has unrivaled access to driver data to validate an automatic driving system. Furthermore DiDi is motivated to develop its own L4, L5 AV to reduce drivers’ costs, and improve passenger safety. Thus, it should be no surprise that last March, Didi established a lab in Mountain View dedicated to self­driving technology. The DiDi lab now has approximately 100 employees, and in May, the company received permission from the state of California to test its self­ driving cars.

Figure 6. DiDi Lab in Mountain View

The eventual champion of AV industry will be the company that produces a highly reliable system that relies entirely on driver data. Whether the winner will be Baidu’s Apollo or one of its competitors cannot be known at this time. However, given Baidu’s mapping expertise, government support and impressive partnership network, its reasonable to expect that Apollo can withstand the competition.

If Apollo does succeed, AV tech companies that participate in the build-out of the Apollo platform will likely be rewarded for their early efforts. In particular, nTheta expects the following opportunities may be available for foreign and domestic AV tech companies:

  • The hardware that Apollo supports should inspire development of the automatic driving industrial chain. AV sensor manufacturers, including LiDAR, radar and camera manufactures, have clear opportunities to participate in the Apollo platform. As the platform continues to evolve and develop new capabilities, other hardware opportunities will undoubtedly arise.
  • AV development is expected to go hand-in-hand with electric vehicle (EV) development in China. Accordingly, battery manufacturers that can improve, battery capacity, charging speed and battery life may benefit from joining the Apollo platform. Likewise, EV motor manufacturers with improvements of torque, size, weight, and reliability should consider joining Apollo.
  • Finally, all AV tech startups in the Apollo partnership alliance may be able to obtain strategic financing from Baidu . Baidu is dedicating significant capital to autonomous vehicle startups. Apollo Fund was founded in Sept 2017. It is dedicated to financing 100 autonomous driving projects over the next three years. One of its first investments was in Smarter Eye, a software developer focused on ADAS (Advanced Driver Assistance Systems). Baidu led the company’s $15M financing in November. As a matter of fact, Baidu has also invested in startups developing autonomous vehicle hardware, such as Hesai Tech[4], iDriverPlus[5], Velodyne[6], NIO[7].

Final Thoughts

Foreign companies considering joining the Apollo Partnership Network are well advised to take certain preparatory steps. First, IP protection will be critical to protecting your technology in this open platform. Therefore, your first step should be to thoroughly review and, if possible, improve your China IP portfolio before exposing your technology to the China market. Second, review your product roadmap and honestly evaluate whether you will be in a position to establish a leading position in your technology niche in China over the next 5 years. Third, conduct a complete review of the relevant Chinese regulations to understand how best to structure your operations when you enter the China market. Finally, undertake the necessary market research to better understand the Chinese automobile and AV market, especially from the perspective of networks and people in the industry. Certain local companies may already have the manufacturing capabilities and authorized supplier relationships that you will need to succeed. A little groundwork and the right connections can save a lot of effort and headaches down the road.

Reference

Leave a Reply

Your email address will not be published. Required fields are marked *