Integrating multiple technologies such as artificial intelligence, communications, semiconductors, and automobiles, autonomous driving involves a long industrial chain and huge value creation space. It has become a battleground for cross-border, competition and cooperation between the automobile industry and the technology industry in various countries. Research firm Mike Consulting predicts that the market for autonomous driving systems will grow to 57 billion U.S. dollars by 2030, and the market size will increase to more than three times by 2040.

Autonomous driving and ADAS market segments are undergoing transformation, with complex requirements for computing and sensing capabilities. Intel believes that there will be three main challenges in supporting autonomous vehicles: performance and efficiency, real-time processing, and safety. In order to maintain a leading edge in a highly competitive environment, automotive system design engineers must design the most appropriate computing architecture. FPGA has unique advantages unmatched by other chip solutions, and is the best choice to meet the ever-changing requirements of the autonomous driving industry.

I/O hub and sensor intake

As the level of driving automation evolves from ADAS L1 to fully automated L5, the number of sensors required and the demand for sensor data processing will increase exponentially. Increasing the number of sensors helps to achieve a comprehensive 3D perception of the surrounding environment to achieve the dual purpose of safety and convenience. In addition, the use of image sensors with higher resolution, pixel depth and frame rate requires the use of multiple communication interfaces and higher data bandwidth .

Intel® FPGAs can provide an ideal solution to meet the system's requirements for flexible IO and high data rates. FPGA can collect data from multiple sensors (with different interface types, data rates, etc.), and convert them into a unified format (such as MIPI CSI-2) for output to calculation elements, and further transmission to the AD system.

Lidar

For AD applications, lidar sensor units are ubiquitous. From basic signal processing in lidar sensors at the edge to advanced features such as fusion and machine learning, a variety of different architectures can be implemented in lidar units. Intel® FPGAs provide excellent flexibility and scalability to help implement signal processing, data fusion, and complex parallel processing tasks to meet the requirements of such applications.

Security Gateway and Functional Safety

FPGAs can manage the safety data processing strategies of autonomous driving systems, and they can be reconfigured to adapt to changing requirements. If you want to build your own security function, you can take full advantage of the latest features of FPGA, including secure boot, secure key storage, encryption acceleration, security lock, unique device ID, secure debugging, and physical tampering detection and protection.