V2X technology will achieve L4 autonomous driving Baidu releases Apollo Air plan

On May 13, Baidu and Tsinghua University's Intelligent Industry Research Institute (AIR) jointly launched the Apollo Air project. This is the world's first and only vehicle-road collaboration technology that uses pure roadside perception capabilities (V2X) to truly achieve L4 level autonomous driving closed loop on open road continuous road networks. It is the highest technical capability in the field of vehicle-road collaboration and will be a major technological breakthrough in the field of global smart travel.

In general, the Apollo Air project has three major features: relying on pure roadside perception (V2X) to achieve autonomous driving, continuous dimensionality reduction to feed back vehicle-road collaborative products, and open source standards to achieve industry sharing.

According to Dr. Tao Ji, general manager of Baidu's intelligent transportation product development, the innovation of Apollo Air technology lies in that it can achieve L4 autonomous driving with vehicle-road-cloud collaboration by using wireless communication technologies such as V2X and 5G without using on-board sensors and relying only on lightweight roadside perception. This is equivalent to Apollo Air technology replacing the perception system of a stable autonomous driving system and achieving autonomous driving through vehicle-road-cloud cooperation. Compared with single-vehicle intelligence, Apollo Air technology is more complex and has a longer system chain. By empowering road-side network products with Apollo Air technology, a car with limited computing power and no on-board sensing equipment can also achieve some high-level autonomous driving capabilities on this road section, which is equivalent to upgrading a manned car to have some unmanned vehicle capabilities.

After continuous research and repeated testing, Baidu's Intelligent Transportation Team and Tsinghua University Institute of Intelligent Industry have completed the intelligent transformation of several intersections in Beijing Yizhuang, Guangzhou Huangpu, Cangzhou, etc., and realized the testing of Apollo Air's pure roadside perception technology in L4 real-world scenarios. By continuously polishing and iterating Apollo Air's pure roadside perception technology, Baidu will continue to feed back existing intelligent intersection solutions, release technology dimensionality reduction to mass-produced vehicle-road collaboration products, and provide highly reliable roadside perception data for shared unmanned vehicle operations and high-level assisted driving.

As shown in the figure above, when a vehicle turns at an intersection, its vision is easily blocked by large vehicles, resulting in a visual blind spot. The beyond-visual-range perception of roadside equipment can push real-time information to the vehicle, allowing it to pass safely.

In the next step, Baidu and Tsinghua Institute of Intelligent Industry will regularly disclose relevant R&D cases and data of Apollo Air through open source and standardized methods, continuously clarify the infrastructure technical conditions that meet the needs of autonomous driving, and build Apollo Air into a shared vehicle-road collaborative technology platform in the industry.

Zhang Yaqin, Dean of the Institute of Intelligent Industries at Tsinghua University, said: The Institute of Intelligent Industries at Tsinghua University is committed to breaking through the core technologies of artificial intelligence and empowering industries. The Apollo Air project jointly developed with Baidu is an exploration of the uncharted territory of China's vehicle-road collaboration technology. In the future, we will have deeper cooperation in technology research and development, talent training and other fields, promote the "university + enterprise" innovation dual engine, and provide more momentum for the development of China's vehicle-road collaboration and intelligent transportation.

Single-vehicle intelligence is the foundation of autonomous driving, but its perception ability cannot completely solve the long-tail problem of complex scenarios. At the same time, the production cost of autonomous driving vehicles remains high in the short term and is not enough to support the commercial scale production of autonomous driving vehicles. Therefore, single-vehicle intelligence + vehicle-road collaboration is the optimal solution for the rapid development of China's autonomous driving industry. The implementation of autonomous driving must rely on both single-vehicle intelligence and vehicle-road collaboration.

As shown in the figure above, in severe weather conditions, vehicle visibility is limited, and vehicle-road collaboration can synchronize road environment information in real time, greatly reducing the occurrence of safety accidents.

At the same time, vehicle-road collaboration technology needs to have extremely high stability and reliability. In the existing vehicle-road collaboration technology route, vehicle-mounted perception is often the main focus, supplemented by roadside perception. In the case of strong vehicle-mounted perception, possible problems and defects in roadside perception cannot be fully exposed and quickly improved. This is also the reason why Baidu and Tsinghua Institute of Intelligence jointly launched the Apollo Air project, which provides more reliable safety redundancy for autonomous driving by building the highest technical capabilities of vehicle-road collaboration.

In recent years, many cities in China have been committed to building pilot demonstration areas for vehicle-road collaboration and Internet of Vehicles. Recently, the Ministry of Industry and Information Technology also stated that it will adhere to the strategic positioning of "dual-wheel drive and coordinated development" of single-vehicle intelligence and network empowerment, strengthen communication and coordination among relevant departments, promote the application of core technology research, and increase the construction of networked infrastructure. Baidu has also been deeply engaged in the field of vehicle-road collaboration. In the "ACE Intelligent Traffic Engine" first released by Baidu in April 2020, vehicle-road collaboration was a unique highlight. Based on vehicle-road collaboration and single-vehicle intelligence, Baidu has been continuously testing and polishing on open roads in more than 30 cities, with a total test mileage of more than 10 million kilometers. In December last year, the Institute of Intelligent Industry of Tsinghua University reached a cooperation with Baidu to establish a joint research center to carry out innovative research on autonomous driving. The Apollo Air plan is also the first appearance of the research results of both parties. In the future, the two sides will continue to work together to bring more digital infrastructure solutions and practices to the transformation of urban transportation and promote the development of the global smart travel field.