Reassessing Baidu’s Smart Transportation Ambitions

Author: Wang Deqing

Editor | Wang Gang

The intelligent transformation of transportation has become an inevitable battle.

At present, the number of motor vehicles in China has risen to 380 million, and the number of drivers has reached 465 million, which means that urban traffic management will become more severe.

However, with the rapid implementation of high-tech technologies such as artificial intelligence, big data, and 5G, the industry has seen the great potential of emerging technologies to subvert the traditional transportation industry. Major manufacturers have entered the intelligent transportation industry one after another, and their strategies are endless.

Among the many participants, Baidu is like a dark horse and has entered the field of intelligent transportation.

In April 2020, Baidu officially released its intelligent transportation solution - ACE Intelligent Transportation Engine, followed by projects across the country; in October, Baidu acquired a controlling stake in the veteran transportation company Hualuyi for 730 million yuan; in November, Baidu ACE Intelligent Transportation Engine won the "World Internet Leading Scientific and Technological Achievement Award"; at the end of 2020, Baidu's annual orders for intelligent transportation exceeded 1 billion; in February 2021, Baidu's world's first autonomous driving MaaS platform serving diversified travel landed in Guangzhou;

Recently, Baidu also released the "Baidu Apollo Intelligent Traffic White Paper-ACE Intelligent Traffic Engine 2.0", bringing a heavily upgraded version of the ACE Intelligent Traffic Engine 2.0.

It can be said that every move made by Baidu in recent years has attracted attention from the industry.

Out of curiosity about Baidu's intelligent transportation business, the author of Leifeng.com had a conversation with the core personnel of Baidu's intelligent transportation to gain an in-depth understanding of how Baidu can break the shackles and make transportation no longer an obstacle to urban development.

In the field of integration of artificial intelligence and transportation industry, Baidu Apollo is probably better known as the leader in autonomous driving with 8 years of experience.

Since 2013, Baidu Apollo has begun to invest in the research and development of autonomous driving technology, with the goal of liberating human drivers in more driving scenarios.

It is understood that Baidu Apollo has been conducting open road tests in nearly 30 cities around the world, with a total test mileage of more than 14 million kilometers. It is one of the few autonomous driving players in the world that can compete with Google Waymo.

"But China's autonomous driving route is different from that of Europe and the United States. Europe and the United States emphasize more on the intelligence of a single vehicle, while China is taking the path of vehicle-road collaboration. There are both smart cars and smart roads. The collaboration of the two can better solve the problem of autonomous driving, and cars do not necessarily need to be so expensive. Naturally, Baidu began to explore some vehicle-road collaboration in 2016." said a core member of Baidu's intelligent transportation.

If autonomous driving is considered an important element of future transportation, then Baidu's entry into intelligent transportation seems to be a natural thing.

From autonomous driving and vehicle-road collaboration to intelligent transportation, does Baidu Apollo have a better solution?

"Traffic problems have not changed, but different eras have different solutions." A core member of Baidu's intelligent transportation told Leifeng.com. He gave an example, taking the simplest traffic light scenario as an example, traditional traffic solutions often involve large-scale, indiscriminate control and guidance of vehicles, not to mention interaction at the level of individual vehicles. However, in the intelligent transportation scenario, the coordination between vehicles and roads can enable road-side information to reach individual vehicles, realize vehicle-road information exchange, and thus achieve more refined traffic management and control.

However, this kind of sophisticated vehicle-road interaction will place very high demands on the C-end reach and cloud service capabilities of transportation players. However, for Internet technology companies, cloud services and C-end capabilities are a natural barrier.

It is understood that products such as Baidu Maps, in-vehicle OS, and smart rear-view mirrors already have a fairly high market share in the automotive market. Through these rich terminals, personalized information can be pushed to individual vehicles, enabling refined management, control, and induction of traffic.

(ACE Intelligent Traffic Engine 1.0 Architecture)

(ACE Intelligent Traffic Engine 2.0 Architecture)

Judging from Baidu's latest intelligent transportation solution - ACE Intelligent Transportation Engine 2.0 architecture, Baidu has upgraded the original "1+2+N" overall architecture to "1+3+N", and added a "Mass Travel Service" intelligent engine based on "autonomous driving" and "vehicle-road collaboration", incorporating "mobility as a service" into the full-scenario intelligent transportation solution.

But whether it is the 1.0 architecture or the 2.0 architecture, the core advantages of Baidu's intelligent transportation are inseparable from four elements: car, road, cloud, and map.

As for cars, Baidu Apollo currently has more than 70 partner brands, more than 300 top ecological partners, and more than 650 models on the market, which can serve tens of millions of vehicle users. Baidu's user scale exceeded 1 billion in 2019, and Baidu Maps, with 400 million monthly active users, is also reaching more users.

At the same time, Baidu has also established a fleet of autonomous driving Robotaxi vehicles with a number of more than 500. These autonomous driving terminals can also help Baidu initially build a traffic data collection system framework.

As for the smart "roads", Baidu deploys intelligent sensing devices on the roadside and at intersections to collect various information such as road traffic, bad weather, and safety facilities, and transmits them to travelers in a timely manner. At the same time, emergency information obtained by the vehicle side can also be fed back to the intelligent roadside sensing for coordination. Through the powerful information exchange capability of vehicle-road collaboration, it is convenient for traffic managers to obtain the latest traffic conditions in a large range and in a timely manner, reducing management costs.

Cloud represents Baidu Smart Cloud. As an AI-defined cloud, Baidu Smart Cloud can provide cloud and edge computing power for roadside equipment.

In terms of "maps", the high-precision maps, three-dimensional maps, and smart maps provided by Baidu Maps, combined with the perception and data analysis capabilities of Baidu's intelligent patrol vehicles and vehicle-road collaborative roadside equipment, deeply integrate map scenes and traffic management, and realize the full life cycle management of equipment, thereby ensuring the stable and healthy operation of the smart transportation system.

It is not difficult to see that Baidu is using AI technology as a medium to explore the potential of intelligent transformation of transportation.

However, there are also some voices in the industry that believe that autonomous driving and vehicle-road collaboration are more like the rising stars of intelligent transportation, and that changes in the transportation industry require more practical implementation methods.

In fact, Baidu insiders, including Robin Li (Baidu Chairman and CEO), clearly realize that it will take time to completely solve the problems in the field of transportation. Therefore, at the beginning of the design of intelligent transportation solutions, Baidu clearly realized that intelligent transportation should not only solve future problems, but also solve the actual transportation problems of current urban development.

The two interviewees from Baidu also agreed that Baidu's intelligent transportation solution can not only solve problems tomorrow and the day after tomorrow, but can also be used to solve very real traffic problems at present, such as traffic safety and congestion.

Take the management of traffic intersections as an example. Traffic intersections have always been the top priority of urban traffic management and the lifeline of urban traffic. If Baidu wants to solve the current traffic problems, then the management of traffic intersections is also an unavoidable problem.

To this end, in March this year, Baidu released the ACE smart intersection solution, hoping to redefine the Chinese standard for new infrastructure traffic intersections with the three major features of "multiple poles in one, multiple sensors in one", "future-oriented and compatible with the present", and "one-time investment with continuous returns". Currently, the solution has been implemented on a 12.1-kilometer road in Yizhuang, Beijing.

Specifically, the equipment deployed at the intersection consists of cameras + lidar + signal light collector + communication equipment + edge computing unit. The edge computing unit component is the core resource product of Apollo, which has the characteristics of high performance and strong computing power, and can be compatible and reused in multiple scenarios to meet the needs of different traffic scenarios.

Based on Baidu's vehicle-road cloud map capabilities, it can achieve four major features: full-factor monitoring, multi-system integration, high reliability, and high openness and compatibility. Specifically:

• Full-factor monitoring: By integrating electronic police, bayonet cameras, radars, fisheye cameras, edge computing and other sensing and computing devices, it can realize the full-factor perception of road vehicles, roads, pedestrians, environment, traffic events, etc.

• Multi-system integration: It can deeply connect the autonomous driving system, urban management system and vehicle network application ecology at the intersection, providing travelers with all-round and multi-angle traffic information;

• High reliability: It can perform edge-side computing on vehicle-road cooperative video, radar, and signal control data, and combine Baidu map data for multi-source data fusion. The core traffic parameter detection accuracy is >97%, which greatly improves system reliability.

• Highly open and compatible: fully reuse existing monitoring and radar products at intersections to avoid duplication of road-side hardware equipment and help beautify the municipal environment.

From environmental perception, road-side brain, product reliability, to integration with existing intersection products, Baidu's ACE smart intersection solution takes into account both the solution to current traffic pain points and the needs of future intelligent upgrades.

It can be seen that while Baidu is entering the field of intelligent transportation with emerging technologies such as autonomous driving and vehicle-road collaboration, it is also making more traffic scenarios equip with AI capabilities.

It is understood that in the "Baidu Apollo Intelligent Transportation White Paper-ACE Intelligent Transportation Engine 2.0", Baidu has also polished out four major solutions: intelligent networking, smart traffic management, smart highways, and smart parking.

But no matter which plan is implemented, it is different from the traditional one-size-fits-all infrastructure construction in the past, but a more long-term intelligent transportation operator model.

Baidu believes that after deploying facilities at urban intersections, how to improve the stability of the facilities and ensure the continuity of services will be issues that future smart transportation players will need to face, and smart transportation operators can shoulder this responsibility.

The intelligent transportation operator model will change the business model of traditional intelligent transportation construction from a one-time integrator model to a continuous operator model. By continuously upgrading the intelligent vehicle-road algorithms that fit the actual scenarios, it will improve the efficiency of urban travel and solve problems that could not be solved by manpower and traditional infrastructure in the past.

With these scene AI capabilities and smart transportation operator concepts, Baidu's smart transportation solutions that are both future-oriented and compatible with the present have been implemented in nearly 30 cities including Beijing, Guangzhou, Shanghai, Chongqing, Baoding, Yangquan, Hefei, Chongqing, Nanjing, and Yinchuan.

Looking back at 2020, Baidu's urban intelligent transportation and vehicle-road collaboration market end-user orders exceeded 1 billion yuan throughout the year. Leifeng.com noted that among Baidu's intelligent transportation orders, vehicle-road collaboration urban transportation orders accounted for 70%.

what does that mean?

Core personnel of Baidu Intelligent Transportation said that this figure shows that the current government has generated great demand and interest in the construction of vehicle-road collaboration.

But fundamentally, the reason why Baidu has been able to gain recognition from more cities is that Baidu provides the government with a safety net in emerging projects such as vehicle-road collaboration.

In fact, due to cost, policy, and regulatory issues, VIS technology has not been widely implemented and applied at present, and it will take some time for VIS to be implemented on a large scale. Therefore, as the first player to try it out, Baidu needs to have more technical and engineering capabilities.

"After all, as an emerging industry, vehicle-road collaboration actually needs some benchmarks with more perceptible effects in the early stages of industry development. So we must produce results and create greater value," said a core member of Baidu's intelligent transportation.

He said that Baidu's products and technologies are not traditional integrator products, but have many technological innovations. For the implementation of these technological innovation products, choosing an external supplier team to implement them is definitely more cost-effective, but there are also certain delivery risks and the delivery and implementation effects are difficult to guarantee.

Therefore, in projects such as intelligent networking and vehicle-road collaboration, Baidu chose the role of technical general contractor. In 2020, Baidu made up for the lack of delivery capabilities of Internet technology companies in the transportation field by acquiring a 730 million stake in Hualu Yiyun.

Not only does it undertake the general contracting, but it also has its own delivery team. It is understood that the size of Baidu's internal delivery team is constantly expanding, with more than 1,000 people.

Building your own delivery team means providing a comprehensive guarantee for the project in terms of plan, technology, and implementation.

Taking the Beijing Yizhuang project as an example, Beijing Yizhuang's networked cloud-controlled high-level autonomous driving demonstration project covers multiple application solutions such as smart traffic management and intelligent networking. It not only provides data support services for current vehicle-road collaboration and smart traffic management applications, but also supports the large-scale operation of L4 autonomous driving vehicles.

At the same time, the integration of multiple poles and multiple sensors requires that traffic equipment be installed on the same pole, and the number of various cameras installed has dropped by more than half year-on-year. More importantly, the equipment on the side of Yizhuang Road can be optimized through OTA, with continuous iteration capabilities.

The project involves enormous technical difficulties and complex chains.

It can be seen from this that Baidu’s goal is not to become a project engineering company, and building its own delivery team seems to be a move that it has to do out of necessity.

"Become an expert first, and then make the product replicable."

However, the large-scale development of intelligent transportation cannot rely on the efforts of a single company. The power of the ecosystem is often the source of the continuous growth of an industry.

To this end, Baidu has been inviting many partners to participate in the Smart Transportation Partner Conference this year in 18 cities including Beijing, Shanghai, Guangzhou, Chongqing, etc. The purpose is to promote the industry with more partners after completing the closed loop of its own technology and delivery.

In general, Baidu has chosen to walk on two legs in the implementation of smart transportation projects. On the one hand, it relies on its own team to deliver, constantly polishing solutions and creating benchmark cases while reducing various costs; on the other hand, it continues to attract ecological partners. Walking on two legs will be the key to the large-scale implementation of vehicle-road collaboration from demonstration areas.

In Baidu's view, ACE starts with transportation, but it is not limited to transportation.

Intelligent transportation construction can actually play a greater role in the construction of smart cities.

First, infrastructure reuse. The real-time feedback data from the roadside is not only needed by the traffic management department, but also by the city's housing and construction, urban management, municipal administration, environmental protection and other commissions and bureaus. If a set of equipment can be shared by various commissions and bureaus, its value will be amplified while the cost will be further shared.

On this basis, the ACE intelligent traffic engine can evolve into a solution for the entire smart city, providing the city with a complete set of intelligent governance systems and systems, thereby bringing greater value.

This reminds me of the "Pilot Work Deployment for the Collaborative Development of Smart City Infrastructure and Intelligent Connected Vehicles" jointly promoted by the Ministry of Housing and Urban-Rural Development and the Ministry of Industry and Information Technology in May this year. Xin Guobin, member of the Party Leadership Group and Vice Minister of the Ministry of Industry and Information Technology, specifically pointed out: "We must strengthen the coordination of pilot demonstrations, deepen the testing and demonstration of intelligent connected vehicles, strengthen the construction of intelligent infrastructure and the ' Car City Network' platform, and provide more diverse testing and application scenarios."

This development path also coincides with the development direction of Baidu ACE intelligent transportation.

As a pioneer, Baidu is constantly implementing technologies to promote the further digitalization, networking and automation of my country's transportation and provide residents with a good travel experience.