At HUAWEI CONNECT 2018, Huawei launched its Mobile Data Center (MDC). Covering chips, platform, operating system, and development framework, the software and hardware integrated solution is positioned to fuel autonomous driving.
Joy Huang, Vice President of Huawei’s IT Product Line, delivered a keynote speech “Powering Intelligent Enterprises with AI Innovation”, said, “In the next decade, autonomous driving is bound to change the human lifestyle profoundly, and innovation will be the driving force behind this change. An autonomous car needs to aggregate and process massive amounts of sensor data to support decision-making for automatic driving. An autonomous car is literally a Mobile Data Center. However, for the century-old automobile industry, also for the ICT industry, autonomous driving is never an easy task. It comes with a highly complex engineering system that involves the bottom-layer chip, platform, in-vehicle operating system, and the upper-layer development framework for running algorithms. Huawei strives to become an enabler of autonomous driving and looks to join customers in the automotive industry to bring the vision of autonomous driving faster to life through continuous innovation.”
The MDC released at the site is the Huawei MDC 600. Compared with current autonomous driving computing platforms in the industry, Huawei MDC boasts the following edges:
Running on Huawei Ascend chips, it unlocks up to 352 trillion operations per second (TOPS) and meets the requirements of L4 automatic driving. High compute power means that it can be connected to more external sensors (such as cameras, millimeter-wave radars, laser radars, and GPS) and process data streams in real time. This will enable the auto-driving car to better cope with complex road conditions in a more prompt, secure, and reliable manner.
High Safety & Reliability
The MDC draws on Huawei’s 30 years of experience and expertise in the R& D, design, and manufacturing of ICT equipment. End-to-end redundancy backup design eliminates single points of failure. It runs stably at -40°C to +85°C ambient temperature, confidently coping with the harsh vehicle environments. It complies with automotive-grade requirements for reliability and function safety, such as the ASIL D level of the ISO 26262 standard.
High Energy Efficiency
Huawei MDC boasts a superior end-to-end energy efficiency of 1 TOPS/W, which leads the industry average of 0.6 TOPS/W. The benefits of high energy efficiency not only mean energy saving, but also stronger endurance of the vehicle battery for more mileage. Also, providing the same compute power at lower temperature means improved reliability for electronic components. In addition, Huawei MDC does not need high-vulnerability components such as fans. This allows for a smaller size of the product and mitigates impact on the existing structure design of cars.
Guaranteed Low Latency
The underlying hardware platform is equipped with a robot operating system (ROS), and the underlying hardware and software are effectively integrated and tuned to achieve a kernel scheduling latency less than 10 μs, and inter-node communication latency less than 1 ms for the ROS. These combine to deliver an end-to-end latency less than 200 ms for autonomous driving by far outperforming the industry’s general 400–500 ms latency. These benefits translate into higher safety for autonomous driving. Huawei has been working in the ICT field for many years. In the intelligent transformation of vehicles, Huawei has cultivated chip-level innovation capabilities, platform-level engineering capabilities, and a comprehensive development, debugging, and diagnosis tool chain. Huawei is poised to work with automobile industry customers to get autonomous driving onto a fast track.