degree of intelligent driving and the importance of chip development

degree of intelligent driving and the importance of chip development

The intelligent level of automobiles has higher and higher requirements for chips

Taking the required computing power as an example, experts in related fields said in an interview with the semiconductor industry watch and other media earlier that if we want the intelligent driving of the car to be close to the L4 or L5 level, the chip requires at least 1000TOPS of computing power. .
And further pointed out that entering the era of intelligent computing including autonomous driving, what we need is not only the computing power of the chip itself, but the close combination of hardware, software and algorithms. Only then can the real computing power that is ultimately required be obtained.
In past developments, the chip industry has been looking to achieve better control over power consumption and area while chips provide performance. Talented engineers can balance these points well in the early stages of process development. We only rely on a millimeter-wave radar or a monocular camera to complete a most basic L1 or L2 level assisted driving.

But now, we have evolved to a variety of sensors (including high-precision radar and high-resolution cameras) that greatly enrich and enhance the amount and type of data we capture in terminal testing; on the other hand, what autonomous driving can do, covering The application scenarios of AI are also expanding, and the scenarios are becoming more and more complex; and with the implementation of advanced autonomous driving, large-scale parallel computing AI computing has also formed a trend. Machine learning represented by neural networks, coupled with the higher-level, richer, more scene-covering, and more sensor-connected advanced assisted driving solutions we mentioned above continue to land, which makes our computing demands continue to grow. promote.

It is driven by these multiple factors that autonomous driving has gradually accelerated the migration from traditional rule-based computing to data-based. “All in all, AI computing will gradually replace logical computing and become the core of in-vehicle computing. The new technology paradigm of autonomous driving needs to match a new computing architecture, making the whole machine more autonomous, making development easier, and making computing smarter.

Share this post