How Data Drives the Future of Intelligent Cars: Insights from Industry Leaders

Introduction

Intelligent vehicles, led by new‑energy cars, have become a flagship of the automotive industry. The exponential growth of car‑network data, higher software engineering demands, shortened product cycles, and strict data‑privacy obligations create significant challenges for manufacturers, software developers, and cloud providers.

Event Overview

On November 2, Tencent Cloud TVP hosted a closed‑door industry meeting titled “Data‑Driven Intelligent Car Development.” Speakers included Fan Haoyu, senior vice president of Li Auto; Li Shengbo, associate dean of Tsinghua University’s School of Vehicle and Transportation; and Liu Shuqian, vice president of Tencent Smart Mobility.

Data‑Driven Product Iteration (Li Auto)

Fan Haoyu explained how Li Auto uses vehicle‑usage data from the Li ONE model to guide product iteration. Family‑oriented travel patterns, increased in‑car entertainment, and voice interaction usage inform feature development, aiming to create a more immersive cabin experience.

He traced AI progress from the 2016‑2017 AI era, through voice and graphics in 2018, deep understanding in 2019, to “AIG” in 2020‑2021, highlighting how these advances shape autonomous‑driving directions.

Li Auto’s autonomous‑driving roadmap focuses on self‑learning and algorithmic upgrades to achieve city‑level autonomy. Engineers analyze massive driving data to locate and solve problems, iterating toward a qualitative leap in safety and performance.

Privacy protection is emphasized through on‑board local computing, dual‑encryption chips (TEE + HSE), selective data authorization (one‑time, yearly, or reject), and features such as encrypted Bluetooth call logs and masked parking photos, earning strong user trust.

General Optimal‑Control Problem Solver (GOPS) for Autonomous Driving (Tsinghua)

Li Shengbo introduced GOPS, a General Optimal‑Control Problem Solver developed since 2018. GOPS is data‑driven, leverages neural networks and brain‑inspired learning, and targets high‑precision, highly generalized intelligent control for complex nonlinear systems such as autonomous driving and rocket attitude control.

The toolchain streamlines traditional control design by separating offline solving from online deployment, improving real‑time performance. It covers the full workflow: problem modeling, network training, simulation verification, code generation, and hardware‑in‑the‑loop testing.

Case studies demonstrated reinforcement‑learning value‑distribution algorithms for multi‑lane autonomous driving, brain‑like decision making at intersections, and other typical scenarios.

Car‑Cloud Integration and New Productivity (Tencent)

Liu Shuqian highlighted three major challenges for the smart‑car industry:

Building a globally integrated infrastructure to support worldwide R&D and comply with diverse data‑protection regulations.

Creating China‑specific products and services, leveraging local maps, ecosystem, and compliance solutions.

Continuously improving user experience throughout the vehicle’s lifecycle, using Tencent’s cloud base, map services, AI, and customizable digital humans. Tencent’s cloud offers dedicated CDZ nodes, unified security environments, and scalable global services to address these challenges.

Conclusion

The transition from engineering‑driven to data‑driven automotive development requires accurate user insight, precise technology trend control, and strict data‑security compliance. TVP’s events aim to disseminate technology, foster collaboration, and accelerate digital transformation across the automotive ecosystem.