Industry News

The core development technology of electric vehicle intelligence

With the continuous improvement and improvement of global technology, the trend of automobile electrification and intelligence is becoming more and more obvious, and semiconductors as the core are becoming more and more important to automobiles. Among them, the intelligence level and intelligence level of the car has gradually become an important factor for customers to consider when purchasing a vehicle. The accuracy and efficiency of autonomous driving depend on the computing power and manufacturing process of the chip. Hence, the increasing demand for autonomous driving is driving the demand for advanced processes and will significantly increase the market size of advanced driver assistance systems (ADAS)/autonomous driving (AD) chips. The computing power of the ADAS/AD processor must meet the requirements of the corresponding autonomous driving level.
The TOPS (trillion operations per second) of Level 2 ADAS/AD chips is generally between 10 and 100, but the TOPS of Level 3 is between 150 and 200, and the TOPS of Level 4/Level 5 is over 400 and will reach 1000+ . Each level is further divided according to function. Basic Level 2 features only include Adaptive Cruise Control (ACC) and Lane Keeping System (LKS), and can be implemented with a SoC as little as 10 TOPS. However, Advanced Level 2 may require up to 75 TOPS for Advanced ACC, which maintains the center of the lane and pre-controls speed on upcoming bends.
The barriers to entry for Tier 1 and 2 ADAS SoCs are low. Therefore, while the cost of ADAS sensors such as cameras and radars continues to decline, ADAS penetration will increase significantly. It is expected that by 2024, the penetration rate of global ADAS in vehicle shipments will reach 78.7%. At the same time, multiple new players will enter the ADAS chip market. These startups have AI chip design and mass production capabilities, and their solutions can quickly meet localized needs at competitive prices.
On the other hand, R&D spending and entry barriers for autonomous driving (AD) chips are significantly higher than ADAS. In addition to core AI chips, AD solutions should include connectivity, sensing systems, image training models, ADAS map development, route planning, vehicle control, driver monitoring systems (DMS), natural language processing (NLP), and smart cockpit solutions . In addition, AD chips must be able to provide customized and region-specific algorithms.