XPeng AD Head Li Liyun Becomes Zhongqing CTO

Post LinkedIn

⚛️Read original on 量子位

#embodied-ai#robotics-leadership#data-flywheelzhongqing-embodied-brainxpengzhongqingli-liyun

💡AD expert leads Zhongqing's embodied brain push with proven tech stack

⚡ 30-Second TL;DR

What Changed

Li Liyun, ex-XPeng AD No.1, joins Zhongqing as CTO.

Why It Matters

Zhongqing gains edge in embodied AI by leveraging proven AD expertise, potentially speeding up humanoid robot development amid fierce competition.

What To Do Next

Benchmark Zhongqing's data flywheel against your embodied AI pipelines for efficiency gains.

Who should care:Developers & AI Engineers

Key Points

•Li Liyun, ex-XPeng AD No.1, joins Zhongqing as CTO.
•Brings AD industrialized AI, data flywheel, engineering to embodied track.
•Targets native multimodal drive with big-small brain and neural collab architecture.

🧠 Deep Insight

AI-generated analysis for this event — not the original article.

🔑 Enhanced Key Takeaways

•Li Liyun previously served as the Head of XPeng's Autonomous Driving Center, where he was instrumental in the development and mass-market deployment of XNGP (XPeng Navigation Guided Pilot).
•Zhongqing (Zhongqing Technology) is positioning itself as a specialized player in the embodied AI sector, focusing on integrating advanced automotive-grade perception and decision-making stacks into robotics hardware.
•The transition of talent from automotive AD (Autonomous Driving) to embodied AI reflects a broader industry trend where companies are leveraging mature 'data flywheel' methodologies—originally perfected for highway/urban driving—to solve generalization challenges in humanoid and industrial robotics.

🛠️ Technical Deep Dive

The 'big-small brain' architecture mentioned refers to a hierarchical computing strategy:

Big Brain: A high-compute, cloud-based or high-performance edge model responsible for complex reasoning, long-horizon planning, and semantic understanding.
Small Brain: A low-latency, real-time model running on local hardware (neural endings) for immediate motor control, obstacle avoidance, and tactile feedback.
Data Flywheel: Implementation of automated data labeling and closed-loop simulation environments to continuously retrain the 'Big Brain' based on edge-case failures captured by the 'Small Brain'.

🔮 Future ImplicationsAI analysis grounded in cited sources

Zhongqing will release a commercial-grade humanoid robot prototype by Q4 2026.

The recruitment of a high-profile automotive AD leader with mass-production experience suggests a shift from R&D to industrialized hardware manufacturing.

The company will adopt a 'software-first' licensing model for its embodied AI stack.

Leveraging Li Liyun's background in scalable software deployment indicates a strategy to provide the 'brain' for third-party robotic hardware manufacturers.