On October 8, Beijing-based embodied intelligence company Lingsheng Technology announced it had closed a Series A round worth more than 100 million yuan, co-invested by Huafang Capital, Daohe Yuanqi under Daohe Changqi, Suwen Electric Energy (300982.SZ), and a leading industrial strategic investor in the embodied intelligence sector.

Founded in February 2023, Lingsheng Technology positions itself as an infrastructure provider for embodied intelligence. The problem it targets is concrete: once robots are deployed in factories, warehouses, and energy facilities, does every switch to a different robot body mean collecting data, training models, and deploying all over again?

The company's answer is RUDA (Robotics Unified Device Architecture), its in-house unified architecture for embodied models. It connects upward to models and applications and downward to robot bodies of different brands and form factors, turning robot capabilities into assets that can be reused, transferred, and continuously improved.

Funding: Three Rounds in a Year, Series A in Under 12 Months

According to Qichacha records, Lingsheng Technology announced two rounds last December, Pre-A and Pre-A+, each worth more than 100 million yuan, backed by Ningbo Fangzheng, Hongruida Investment, and Wanwuwei Venture Capital. The new Series A was co-invested by Huafang Capital, Daohe Yuanqi under Daohe Changqi, Suwen Electric Energy, and a leading industrial strategic investor in embodied intelligence.

From Pre-A to Series A, the company closed three rounds in less than a year.

Technology: RUDA as a Unified, Cross-Body Foundation

Lingsheng Technology's core product is the RUDA unified robotics system architecture, built around a self-evolving loop that runs from data collection to model training, robot execution, and feedback.

The architecture hinges on the RUDA universal Harness system, which supports heterogeneous hardware including humanoid robots, various dexterous hands, and industrial grippers, so upper-layer models and applications do not need rewriting every time the underlying robot changes.

Within the RUDA system, the architecture splits into two agent layers. RUDA serves as the foundation, PhiAgent handles data generation, and RoboAgent handles execution in the field, with execution results fed back for optimization so robot capabilities keep improving.

Division of Labor: Turning Human Video into Robot Skills

PhiAgent is the data and capability production engine of the RUDA system, built to convert videos of human operations into data and skills that robots can reuse. It bridges ordinary hand-motion video to robot simulation, generating robotic arm trajectories and simulation playback, and can produce data tailored to different robot bodies, building an asset chain that links human video, robot video, and simulated trajectories.

RoboAgent is the execution agent for real-world settings, responsible for understanding task goals, calling the relevant skills, controlling the robot through a complete operation, and adjusting on the fly. The two now work in tandem, and the company is extending them to deformable-object tasks such as folding clothes.

To keep the architecture's data loop iterating, Lingsheng Technology also developed LivDEX, an exoskeleton-style glove for multimodal hand data collection.

Deployment: Near-100% Success Rate on Optical Module Insertion

The company says the RUDA architecture has been deployed across smart manufacturing, unmanned retail, healthcare and elderly care, research and education, and tourism services.

On the commercial side, Lingsheng Technology has deployed in high-precision industrial tasks such as optical module insertion and removal and battery assembly. Optical module insertion reached a near-100% test success rate under specific conditions. The company also covers 3C quality inspection, in-plant material transfer, and large-component handling, and has delivered an integrated cloud-edge-device solution.

Team: A Post-2000 Tsinghua PhD as CTO, a Tenured Tsinghua Professor as Adviser

Founder and CEO Yang Hongbing is a Tsinghua University graduate with a long track record in large models and robotics. He previously served as senior algorithm director at Baidu and head of algorithms for its on-device AI engine, then as a vice president at Tencent Cloud and head of its commercial large-model platform product line, and as CTO of a humanoid robot unicorn.

Co-founder and CTO Jiang Yuhua, born after 2000, holds a PhD from Tsinghua University. He studied in Tsinghua's Qian Xuesen Class as an undergraduate and completed his doctorate in the Department of Automation, winning Tsinghua's top scholarship and the IEEE WCNC 2024 Best Paper Award. He leads the company's core technology strategy, its full-stack R&D system, and the technical roadmap for the world-model Harness, focusing on physical AI infrastructure and self-evolving loops for robot data and models.

Gao Feifei, a tenured full professor in Tsinghua University's Department of Automation, serves as the company's science and technology adviser. Gao is a recipient of the National Science Fund for Distinguished Young Scholars and an IEEE Fellow, has won a first prize for natural science from the Chinese Association of Automation, and has been named a Clarivate Highly Cited Researcher multiple times, with nearly 30,000 total citations.