At the Qiankun Intelligent Technology Conference on April 22, Huawei unveiled a series of smart automotive solutions, with the centerpiece being the latest edition of the Qiankun autonomous driving system (ADS), version 4.0. This modular platform comes in four versions, with the “Ultra” model reportedly reaching Level 3 autonomous driving capabilities on highways.
Transitioning from Level 2 to Level 3 autonomy required Huawei to significantly upgrade its hardware, system architecture, and chassis integration. As part of the launch, the company introduced three new types of radar:
- A compact, high-precision solid-state LiDAR (light detection and ranging) sensor, offering three-centimeter accuracy to enhance detection of objects behind the vehicle.
- An in-cabin laser vision sensor that improves the system’s ability to spot small and distant objects, enabling emergency braking at speeds up to 100 kilometers per hour even in low-light conditions.
- A distributed millimeter-wave radar that strengthens detection reliability in adverse weather such as rain, fog, and dust.
Data collected by these sensors are processed by Huawei’s WEWA architecture, running on the MDC 1000 platform powered by the Ascend 910B chip. With computational performance reaching 1,000 TOPS (tera operations per second), the system handles vast streams of sensor data in real time. Integrated with Huawei’s XMC digital chassis engine, the platform reportedly reduces end-to-end latency by 50%, improves traffic efficiency by 20%, and cuts emergency braking incidents by 30%.
WEWA is built around two key components: a cloud-based “world engine” that generates complex virtual driving scenarios, and an in-vehicle “world behavior” model trained to navigate these scenarios. This structure addresses the challenge of edge cases—rare or unpredictable events that traditional autonomous systems often struggle with.
Events such as a pedestrian suddenly crossing the road or malfunctioning traffic lights are hard to capture in real-world datasets. While Huawei’s earlier ADS 3.0 platform used an end-to-end learning model, it was limited by a shortage of edge case data. The updated WEWA framework overcomes this by generating synthetic edge cases in virtual simulations, allowing the system to pre-train for unusual real-world conditions.
Supporting these perception improvements, Huawei’s XMC digital chassis engine integrates driving intelligence and vehicle control into a unified system. By centralizing control processes, XMC is said to improve system responsiveness by a factor of ten and increase the number of coordinated components fivefold compared to traditional designs.
Together, these advancements strengthen the case for deploying Qiankun ADS 4.0 Ultra in real-world Level 3 autonomous driving scenarios. However, industry experts note that no simulation can account for every possible real-world anomaly, and caution remains around the limits of virtual training.
Huawei also revealed HarmonySpace 5, its next-generation smart cockpit platform. The system improves voice interaction success rates to 85% and uses the Yunque artificial intelligence model to diagnose vehicle issues with professional-level precision. For audio, Huawei Sound supports up to 43 speakers, creating a 7.5.10 surround sound experience. A new 16.1-inch display, combined with an image enhancement engine, boosts video resolution from 720p to 1440p.
Additionally, HarmonySpace 5 synchronizes ambient lighting, wallpapers, and screen savers to create an immersive in-cabin atmosphere.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This article was written by Fan Shuqi for 36Kr.