DiDi Launches Voyager Labs for AV Research

Post LinkedIn

🇨🇳Read original on TechNode

#autonomous-driving#world-models#embodied-aivoyager-labsdidivoyager-labstsinghua-universityli-shengbo

💡DiDi's lab pioneers multimodal RL for true end-to-end self-driving tech.

⚡ 30-Second TL;DR

What Changed

Launched DiDi Voyager Labs for autonomous driving R&D

Why It Matters

Advances end-to-end AV tech via multimodal AI, potentially accelerating commercial deployment. Academic ties boost innovation in embodied AI.

What To Do Next

Monitor Voyager Labs papers on arXiv for multimodal world models in AV.

Who should care:Researchers & Academics

Key Points

•Launched DiDi Voyager Labs for autonomous driving R&D
•Targets multimodal LLMs, world models, and RL
•Partners with Tsinghua Prof. Li Shengbo's team
•Joint framework for dedicated research

🧠 Deep Insight

Background and context from public sources — not the original article. 8 sources cited.

🔑 Enhanced Key Takeaways

•DiDi Autonomous Driving was spun off as an independent company in 2019 after starting its research team in 2016, and has secured public road testing permits in multiple Chinese cities and California.[2]
•In October 2025, DiDi Autonomous Driving raised CNY 2 billion (USD 281 million) in Series D funding from investors including GAC Group and Beijing AI funds to advance L4 autonomous driving and AI research.[4][5]
•DiDi partnered with GAC Aion to develop the Robotaxi R2 based on the Aion V SUV, which began real-world public road testing in Guangzhou following 24/7 driverless taxi trials.[3]

🔮 Future ImplicationsAI analysis grounded in cited sources

Voyager Labs will accelerate DiDi's L4 robotaxi commercialization by 2027

The lab's focus on multimodal LLMs, world models, and RL builds on recent Series D funding and Robotaxi R2 testing to integrate research with engineering for end-to-end AV deployment.[2][3][4]

Partnership with Tsinghua will enhance talent pipeline for AV AI

The joint framework includes talent training programs combining academic expertise in AV with DiDi's operational data from public road tests.[2]