(Chinese Version)

Outside the CES Asia’s N3 venue, Baidu publicly demonstrated an autonomous car co-developed with Nvidia and GWM.

It wasn’t exactly a complex display that required much efforts. According to Baidu, the demonstration zone was isolated from the public six weeks before the road test for the car to learn and practice. Powered by terminal-to-terminal and single-lens autonomous driving deep learning solution, the car eventually learned to drive autonomously by itself by making mistakes and practicing.

TMTPost found that the autonomous had been running smoothly without any sudden braking during the whole trip on autonomous driving mode when experiencing the user experience. When there was a new traffic signal, the car would respond accordingly.

As a matter of fact, it’s quite common to see autonomous car road tests nowadays and it isn’t the first time for Baidu to test autonomous cars. In 2015, Baidu had tested its autonomous car’s lane change, deceleration, overtaking functions on the highway and ring road. Last year in Wuzhen, Baidu’s autonomous car division even established Yunxiao fleet, performing an experience campaign in which they ran for several kilometers in the city’s public road.

It’s not easy to go from experiment to road test and mass production. In April this year, Lu Qi announced the Apollo Project. After two months, Baidu provided a grander scope of the project at the CES Asia on providing an open, complete and safe software platform for Baidu’s car and autonomous driving partners.

On the first day of CES Asia, Baidu announced to sign agreement with four companies in an attempt to accelerate the mass production of autonomous cars. The cooperation is based on Badu’s autonomous driving computing platform BCU announced on the same day.

BCU is comprised of two fundamental functions, which are cyber security and cloud update, and three AI core modules, which are precise GPS, environment sensing, and decision planning. At present, Baidu has brought about three products that can fit right into our palm.

BCU-MLOC positions as a computing platform that provides highly precise GPS service while BCU-MLOP serves as an integrated platform that provides both highly precise GPS service and environment sensing services. Besides the GPS and environment sensing services, BCU-MLOP2 on the other hand also provides a computing platform that makes decision.

“BCU is a channel for us to enter the mass production phase. Without it it’s hard for our technologies to enter the market. BCU will be our AI’s channel to enter the automobile industry,” Baidu’s smart car business’s general manager Gu Weihao told TMTPost, describing the computing platform.

On the second day of CES Asia, Gu also mentioned another technology, the Road Hackers, a terminal-to-terminal advanced autonomous driving model.

The autonomous driving technologies mentioned earlier are also based on this model when under development. Its biggest feature is that it allows cars to perform terminal-to-terminal deep learning via a single camera, which Baidu believes would be the future mainstream in the autonomous driving sector. Gu also mentioned that Baidu would make the autonomous training data collected from over ten kilometers’ driving test under the Road Hackers model to accelerate the update of autonomous driving solutions.

In fact, earlier in late May and early June, Baidu had reached a strategic agreement with the Continental Corporation and Bosch. It’s not something uncommon anymore for industry giants to work together nowadays.

This year in March, Intel poured in $15.3 billion to fully acquire Israel’s top autonomous driving solution provider and ADAS provider Mobileye. Within two months after that, BMW, Intel and Mobileye jointly announced to Delphi Automotive, becoming the company’s newest autonomous driving platform development partners and system integrators. The four companies stated that they would adopt a standardized model to expand the scope of autonomous driving solutions for the automobile industry.

It’s worth noting that these giants are going towards the same direction with some other companies that have been mentioned together with them on the news:

On Baidu’s conference last year, Nvidia announced to work with Baidu on developing autonomous driving platform to make a cloud-to-car-architect platform.

On early this year’s Bosch ConnectWorld Conference, Bosch and Nvidia announced that they would co-develop an autonomous driving system based on Nvidia’s DRIVE PX platform and AI chip for cars Xavier.

In earlier years, Benz and Baidu had worked together on developing Internet of vehicles and now the pair has made progress on Benz’s mass-produced cars. Meanwhile Baidu had signed strategic agreement with Bosch.

Seemingly complex, this relationship is actually very easy to understand. The autonomous driving sector is becoming clearer in shape, and it’s unwise to be a lone wolf. Mass production is a project larger than people’s imagination. It’s not something a single company can pull off.

Baidu’s strength lies in its software technologies. For instance, Baidu develops computing platform BCU and optimized precise map. The Internet giant might have troubles if it started to develop hardware. After a century’s development, the automobile industry has formed a rather high industrial threshold. Working with automobile companies will help Baidu make up the disadvantages in hardware and allow Baidu to utilize their influence in the automobile industry. Besides that, Baidu can also improve its lack of car making gene through this channel.

When speaking of the selection of partners, Gu revealed Baidu’s standards: “We are looking for partners that have already had some achievements in the hardware sector and know about the standards in this industry. They should be able to integrate what we provide according to the industrial rules and standards.” Besides the market and capability perspective, Baidu also values the potential collaborators’ plans. “We want partners that can look into the future and not hold on to the current interest.”

Autonomous driving is a brand new model. When autonomous cars can be put into mass production, the autonomous driving industry would have profound impact on rules, regulations and policies.

It’s possible that a parallel structure would appear in the next coming years whereas everybody influences each other while developing. When the situation is more stable, a new product structure would emerge. This industry’s structure will tend to become stable. But whenever something new emerges, the structure will restructure and form a new form. So we need to wait for some time and see what the new industrial structure will be like in the future.

When talking to Gu, TMTPost noticed that he mentioned the phrase low cost and high investment, which comes from Baidu’s consideration on the new cost generated from the mass production of autonomous driving technologies.

For instance, the price of the mainstream automobiles selling in China ranges from ￥100,000 to ￥150,000. If adding an autonomous driving technology, the price would surge to ￥300,000. Nobody would want to pay for that. The issue of cost comes in big when making autonomous a mass production.

More and more giants are working together to develop a standardized computing platform. The incentive here is that they want to be the one that makes the android system in this indusry.

According to Baidu, project Apollo will provide a complete software, hardware, and service solution, which includes obstacle perception, trajectory planning, vehicle control, and operating systems. It’s an open and safe platform open to Baidu’s partners in the autonomous driving and automobile industry, covering vehicle, hardware, software and cloud platform.

“We are not trying to become another Apple. We are uniting forces. In other words, we are forging vertical integration in this industry as well as horizontal cooperative development,” Gu told TMTPost.

However, the question of if there is any environment for lone wolves to thrive in the autonomous driving sector remains. Many might think of Tesla in this sense. And time is the ultimate confirmer.

………………………………………………………………………………………………………
(Like our Facebook page and follow us now on Twitter @tmtpostenglish, on Medium @TMTpost, on Instagram @tmtpost_english and on AppleNews@TMTpost)

The article is published with authorization from the author @Zhang Lin, please note source and hyperlink when reproduce.]

Translated by Garrett Lee (Senior Translator at PAGE TO PAGE), working for TMTpost.