How Close Are We to General-Purpose Robots? | Y·Robot × World Robot Conference

After nearly a week of heated activity in Beijing, the World Robot Conference is winding down. On the afternoon of August 20, at the "2023 Venture Capital Stars — Robot Excellence Project Industry-Finance Collaboration Matchmaking Conference," Yunqi Capital partner Chen Yu participated in a roundtable discussion as a key partner, joining senior experts from academia, industry, research, and government to explore emerging investment trends in "AGI + Robotics."

· Roundtable Guests:

Jianwei Zhang — Academician, German National Academy of Science and Engineering; Professor, University of Hamburg

He Jing — Founding Managing Partner, SEE Fund

Chen Yu — Partner, Yunqi Capital

Qian Zhehong — CEO, Yinniu Microelectronics

Lai Jie — Founder and CEO, Astribot

ChatGPT's explosive debut showed people a path toward artificial intelligence, but models remain fundamentally at the "software level." To bridge the digital and physical worlds, robotics and embodied intelligence are the best vehicles.

Chen Yu noted that R&D priorities in robotics have shifted from the past: no longer limited to completing a single category of task, but advancing toward "multi-type" task execution. In the past, for example, delivery robots handled delivery, construction robots painted walls. But against the backdrop of general-purpose intelligence, future robots may achieve "flexible deployment": a factory robot that can tighten screws, spray paint, and assemble parts.

More highlights from the discussion below:

Jianwei Zhang — Academician, German National Academy of Science and Engineering; Professor, University of Hamburg

Going forward, the challenge of multimodality lies in fusing image processing, olfaction, touch, motor perception, and interactive information. For ChatGPT to be localized in China, it's crucial to find Chinese characteristics, avoid homogenized investment, pursue differentiated development, and capitalize on new opportunities.

He Jing — Founding Managing Partner, SEE Fund

People tend to overestimate change in one year and underestimate transformation over ten. Embodied intelligence is still in early stages of commercialization. We strongly hope that capital, policymakers, and entrepreneurs can collaborate to break down problems and drive implementation, rather than compete blindly. I remain optimistic about the future flourishing of embodied intelligence.

Qian Zhehong — CEO, Yinniu Microelectronics

Large models could give robots a truly intelligent "brain." We're now combining language models and 3D physical world models to train robots to complete tasks from instructions. In the future, robot pre-training may no longer require preset scenes and rules. I believe embodied intelligence will have a profound impact on the robotics industry.

Lai Jie — Founder and CEO, Astribot

Companies face both demand pull and technology push. The greatest demand for embodied intelligence may be the aging population five to ten years out. This year, ChatGPT showed how robots can generalize into universal robots. Meanwhile in China, upstream supply chains, large models, AI teams, and robotics teams are all up and running. I believe helping address aging represents the greatest value of embodied intelligence.

Chen Yu — Partner, Yunqi Capital

Building embodied intelligence requires breakthrough innovation at the intersection of software and hardware, plus close collaboration between industry and academia with sustained investment. We hope the development of AI and robotics can liberate humanity from heavy, repetitive labor — this is the mission that drives our commitment to AGI + robotics.

Additionally, multiple Yunqi Capital portfolio companies are showcasing their latest robots across exhibition zones and forums. We'll continue sharing firsthand updates from the conference, and welcome you to bring your curiosity and interact with the robots.