Baidu is serious about Apollo’s tech and partnerships

Baidu’s Baidu Create: Artificial Intelligence Developer Conference was held at the National Convention Center in Beijing from July 4-5.

During the first day’s event, Baidu made four major announcements. The tech giant said its self-driving minibus will enter mass production phase. New AI chip Kunlun aiming at cloud and edge computing in AI use cases such as autonomous driving and natural language processing can expect more applications. Besides, Baidu unveiled the 3.0 version of key AI project Baidu Brain, one major upgrade of which is Baidu’s voice AI system DuerOS 3.0. The company is also following the trend of mini-programs by leveraging Baidu app.

Baidu’s Apollo Pilot, an L3 level passenger car demo, on display during the event (Image Credit: TechNode/Runhua Zhao)

Beyond announcements of achievements and new plans, the Conference organized forums hoping to push forward research and exhibitions. Technology and partners are becoming Baidu’s major focus to allow the company to turn ambition into reality and secure market profit channels.

International and domestic partners

What’s interesting is that many forums taking place in conference rooms with capacities ranging from around 100 – 500 held Chinese presentations with highly professional (even academic) powerpoint slides fully written in English. While it may seem strange to outsiders, it’s already a very common phenomenon in China’s academic and technology sharing events. The country’s researchers and talents are having increasing international exposures, and China’s catching-up position in world tech realm demands familiarity with English to enter leading research circles. Speakers’ fluent and frequent references to English terms and research literature imply that invited audiences possess related knowledge too.

Meanwhile, entering some forum rooms required scanning of a specific code attached to the registration badge – a procedure not applicable to media attendees who may directly go in.

The signal is clear: the Developer Conference is for developers. Study of and connection with the developer and partner participants are the keys to Baidu’s knowledge and potential building. And the forums were filled with serious-looking developers and manufacturers carefully taking notes and photos of powerpoint slides.

Even during lunch time, carrying a meal just purchased from a convenience store, a man wearing registration badge asked another man walking next to him, “What do you think of the talk on ADAS (Advanced Driver-assistant System)?”

They were probably referring to an autonomous driving presentation held in the morning. William Zhong, Senior Director of China DC at Xilinx, world’s leading semiconductor company offering adaptable solutions, presented the US company’s ADAS insights. Prior to him, Infineon Technologies, a former semiconductor affiliate of Siemens and now the world’s top 3 car semiconductor solution provider, shared general car industry’s major trends namely low-emission, autonomous driving, internet of cars, and information security. The company also introduced Aurix 2G, the technology that assists Baidu’s latest Apollo 3.0 to complete core computation.

Xilinx introducing an autonomous driving solution (Image Credit: TechNode/Runhua Zhao)

In the exhibition area, TechNode noticed Intel’s cooperation with Baidu on Apollo’s driving and route optimization. Besides Baidu, the technology is also used with Mobileye, the autonomous driving ADAS solution company acquired by Intel in August 2017.

Intel’s simulation of driving scenes for the performance and route optimization of Baidu’s Apollo (Image Credit: TechNode/Runhua Zhao)

In terms of hardware, ZF and Nvidia’s logos are also seen on Baidu Apollo’s simplified control box demos.

The simplified control box demo: NVIDIA x Baidu  (Image Credit: TechNode/Runhua Zhao)

In addition to big names, Innovusion, a Silicon Valley-based private company specializing in LiDAR – one of the most crucial and controversial sensors for autonomous driving – presented their “game-changing weapon” solution for the field, and quoted the data research circles’ known rule Garbage in, Garbage out when stressing how quality data have pushed forward their innovation.

Baidu is seizing every opportunity to take any potential pioneer position in the autonomous driving field, and its massive data collected via nationally famous software such as Baidu Map is providing rich raw materials for AI training and pilot projects. Baidu’s global partners expressed excitement when referring to cooperation with this Chinese giant – cooperating with a leading player and resource holder means toeing the starting line to win the game is not always necessary, as your line can be already ahead of others.

Additionally, Baidu’s domestic partners are eagerly showing their achievements. Dr. Xu Huafeng, CEO at StarNeto, a Chinese navigation and infrastructure examination company offering military-level products and solutions, gave a speech on inertial navigation, a technology that relies on few external references when tracking and positioning a moving object. This solution will enable navigation when a target object is covered by trees, bridges, and other objects that may block external navigation signals. The technology is not new. Early in 2007, an inertial navigation Technical Report authored by a current Senior Staff Software Engineer at Google was published by Cambridge University’s Computer Laboratory. Dr. Xu, having addressed the technology’s challenges, remains optimistic about the future of inertial navigation.

According to a participant who wishes to remain anonymous, the Conference was also a chance to expand their own connections, and this will naturally and significantly expand Baidu’s ecosystem and power in the field. He added that highly-professional talks distinguish this event from others. Particularly, learning is more important to insiders than the news that often attracts outsiders’ attention.