Baidu pitches Apollo as open source standard for autonomous vehicles

The US-centric nature of much of the analysis of the connected car industry means it is often portrayed as a showdown between Motor City and Silicon Valley in the US, forgetting that a great deal of the innovation is happening in Asia. It is China’s search giant Baidu which has launched the most comprehensive attempt yet to provide a unified reference platform for autonomous driving.

Of course, given the state of trade and technology rivalry between the US and China, its dreams of global adoption are unlikely to come true. The US is increasingly a large hole in the Chinese commercial map, just as China is for otherwise global firms like Google. But Baidu’s Project Apollo may help to accelerate progress in its own region, and provide some valuable pointers elsewhere.

Project Apollo’s most ambitious aim is to unite the automotive industry around Baidu’s portfolio of autonomous driving technologies – including software platform with advanced machine learning, hardware design and reference vehicles.

Baidu is promising an open software platform for its partners in the autonomous driving industry, so that they can develop their own driving systems based on these reference designs. Baidu believes its “robust, mature, and secure autonomous driving technology” can provide similar potential benefits to those of the Apollo lunar program, from which Baidu borrowed the name to convey its belief in the transformative impact of self-driving vehicles.

Baidu plans to monetize the platform by controlling the API (application programming interface) layers for mapping and machine learning. So while most of the Apollo platform would be free, adopters would pay to access the required upper layer functions.

Trade tensions aside, would the US auto industry get on board, perhaps to weaken the attempt by Google, Tesla, Apple and others to wrest control from Detroit in the next wave of car evolution? But Chinese carmakers are even more direct and threatening competitors, especially with US manufacturers trying to gain share of the domestic market in China.

A similar response from the European auto makers seems likely. Their current approach is to create an inhouse system for self-driving vehicles, before trying to come together as an industry in a formal standardization process. Notably, Baidu and BMW’s development agreement fell apart, with BMW citing different strategies as the cause.

Baidu is right in saying that it is one of the first companies to open up its platform for collaboration, though despite deep pockets (it has a market cap of about $60bn) and a machine learning development to outshine Google’s, it has not made much impact in the next generation auto space yet.

That’s where Apollo comes in. For now, however, Baidu seems far better placed to become the platform of choice for the Chinese auto makers, which lack the budgets to develop an alternative on their own dime. The Western/Indian auto markets are far more consolidated, which is why there are large groups with cash to fund internal projects.

So domestic players look set to be the first adopters for Apollo. Baidu says it will open up the source code for its obstacle perception, trajectory planning, vehicle control, vehicle operating systems, and a complete set of testing tools. These are all technologies that will have been developed in partnership with Chinese car manufacturers like BAIC, BYD and Chery.

Baidu holds a Californian testing permit for its self-driving designs, but has also completed several successful tests in Beijing and Wuzhen. In terms of the roadmap for Apollo, it says it plans to launch its partnership alliance, before opening its tech in a ‘restricted environment’ in July. The next step will see it share its simpler self-driving technology by the end of the year, and then gradually increasing the available complexity – culminating in fully autonomous driving on public roads in 2020.

BAIC and Baidu announced a strategic partnership to develop an SAE Level 3 capable vehicle, back in January, which is due to demonstrate the fruits of their labor at the Shanghai Auto Show, followed by public-road testing throughout 2017. BAIC is providing its some of its OEM solutions to Baidu here, including its CoDriver, MapAuto, and CarLife software – the last of which is apparently used by 700,000 car owners as a smartphone integration system for IVIs.

The other big name involved with Baidu is Nvidia, the maker of graphics processors which has been turning its head towards automotive applications via its software and Drive PX hardware. Announced back in September, the pair said they would work on a cloud platform for Level 3 self-driving vehicles, combining Nvidia’s compute power with Baidu’s mapping technology to create an operating system for these vehicles.

It seems unlikely that some of this technology hasn’t made its way into Apollo, especially as Baidu recently announced that Nvidia’s deep learning platform, based on its Tesla P40 GPUs, was now available to Baidu Cloud customers.

Wi Lu, Baidu’s COO, said in a statement: “China is the world’s largest market for automotive sales and production. It has many car brands and an open environment that is ripe for collaboration. Baidu took the initiative to open our technology to the industry in order to encourage greater innovation and opportunities. An open, innovative industry ecosystem initiated by Baidu will accelerate the development of autonomous driving in the US and other developer automotive markets.”

Meanwhile, Apple’s continues to fuel speculation about the depth of its commitment to the autonomous vehicle market. It has recently received a permit from the Californian driving authorities, allowing it to test three autonomous Lexus SUVs. But a fleet of this size suggests that Apple is using the permit to trial software, rather than aiming for a major self-driving vehicle initiative based on hardware designs – which would be a relief to the established auto makers.

Indeed, the move could mark the company’s exit from the hardware game in this area, just as Google appears to be doing, having recently spun off its Waymo self-driving car project to focus solely on LiDAR sensors and software.

The newly granted permit gives Apple permission to test three 2015 Lexus RX450h SUVs with six drivers. It will be joining another 29 companies with permits to test autonomous vehicle technologies on Californian highways. Major names on the list include Volkswagen, Mercedes-Benz, Google, Tesla Motors, Bosch, Nissan, GM, BMW, Ford, Honda and Baidu.

When Waymo fell out of favor with Google, the move hinted at Google sidelining its autonomous hardware ventures in favor of positioning Android Auto as its major bid to rule the automotive space, via in-vehicle infotainment (IVI) systems. Apple could well follow suit with its CarPlay.

Apple and Google can take advantage of their dominance of the smartphone market, by extending this power, through Android Auto and CarPlay, to bridge the phone and the car, and harness vehicle data to build additional revenue streams . This is the heart of the tensions with the carmakers, which want to control the interface with the driver, and see the telematics and usage data as a revenue stream that is rightfully theirs.

But Apple and Google also face competition in this area from Amazon and its Alexa AI, which has landed integration deals with a handful of major car manufacturers. This could give the car industry a less obviously dangerous party to limit the dominance of Apple and Google – the same instinct which boosted Nokia HERE in mapping (and that is now owned by four carmakers).

Breaking into the automotive industry is no easy feat, but Apple and Google are still keeping their options open to better their chances of getting one or more pieces of the pie – especially as auto players could potentially settle on a de-facto IVI standard in the future, with the development of SmartDeviceLink (SDL) technology.

Apple initially submitted its self-driving car filing to the US NHTSA back in November last year. In that, Apple’s director of product integrity, Steve Kenner, cryptically stated that “the company is investing heavily in the study of machine learning and automation, and is excited about the potential of automated systems in many areas, including transportation.”