Yimu Tech Secures Over RMB 1 Billion in Series E Funding, Valuation Exceeds RMB 10 Billion

Yimu Technology recently closed a Series E financing round exceeding RMB 1 billion, achieving a valuation surpassing RMB 10 billion. The round was jointly led by multiple top-tier RMB funds, leading USD-denominated funds, and prominent strategic investors. Proceeds will be allocated toward R&D in embodied intelligence tactile sensing materials, chips, algorithms, and models, as well as scaling up mass production and ensuring high-quality delivery of manufacturing orders. The company specializes in embodied intelligence tactile perception technology. Its proprietary vision-tactile sensor converts microscopic deformation images at contact surfaces into multidimensional tactile data—including pressure, shear force, and contact geometry—enabling robots to perceive in real time the tightness, slippage tendency, and contact state during manipulation, thereby achieving compliant and disturbance-resistant dexterous operations. Currently, Yimu’s sensor thickness is under 3 mm, making it the world’s thinnest commercially producible bionic vision-tactile sensor, effectively addressing the industry-wide challenge of spatial constraints in dexterous robotic hands. Key performance metrics include: resolution exceeding 10,000 sensing points, force accuracy at the milligram level, force resolution of 0.005 N, industrial-grade durability exceeding one million actuation cycles, and 90% of core tactile performance indicators matching human-level capabilities. Technologically, the company conducts deep in-house R&D across five domains: materials science, precision optics, mechanical simulation, integrated chips, and AI algorithms. It has established a three-layer architecture: - **Layer 1**: Tactile data acquisition solutions and standardized datasets, integrating mainstream data collection techniques such as UMI, DexUMI, and Ego to record force requirements and tactile feedback from real-world interactions. - **Layer 2**: A proprietary Tactile Transformer Encoder algorithm module that unifies raw signals from diverse hardware, extracts information on surface morphology, material hardness, roughness, and other properties, and encodes them into a universal representation. - **Layer 3**: Multimodal embodied intelligence foundation model training, which fuses tactile, visual, and language data to complete the tactile perception loop in manipulation tasks, ultimately producing an end-to-end model capable of perception, decision-making, and action. Yimu Technology’s products are already deployed across diverse sectors including humanoid robotics, automotive, data acquisition, pharmaceuticals, and food processing. The company is collaborating with international institutions such as Stanford University on the TouchNet open-source tactile dataset initiative, aiming to release additional real-world tactile data by the end of 2026 to enable researchers and engineers to access tactile data and test algorithms at low cost.