This post-2000 entrepreneur from Guangdong has already raised 2 billion yuan in financing.

Robot Forward reported on March 10th. Today, Lingchu Intelligence announced the completion of 2 billion yuan in angel and Pre-A round financing.

The angel round investors include China Development Financial, Guozhong Venture Capital, CCTV Media Convergence Industry Investment Fund, as well as two core industrial leading capitals, namely the strategic investment under a listed company worth hundreds of billions and the fund under Yangtze Optical Fibre and Cable Joint Stock Limited Company. The Pre-A round was led by Shanghai State-owned Assets Xuhui Capital, with some local state-owned assets and market-oriented funds following the investment. Many old shareholders made oversubscribed follow-on investments.

Lingchu Intelligence was founded in September 2024, focusing on building a software and data collection toolchain system centered around the end-to-end VLA model. This round of financing will also be invested in Lingchu Intelligence's large-scale application in the logistics scenario and the construction of the large-scale data collection solution system.

Previously, Lingchu Intelligence launched the industry's first end-to-end reinforcement learning embodied model Psi R0, R0.5, and R1, becoming the first enterprise in the industry to develop an embodied model capable of long-range tasks, which also features generalization, robustness, and dexterity.

In December last year, Lingchu Intelligence released the world's first embodied native human data collection solution, Psi-SynEngine. This solution includes a portable exoskeleton tactile glove data collection set, a large-scale in the wild data collection pipeline, and a cross-ontology data migration model based on the world model and reinforcement learning. It has also been the first to apply the collected human data to real scenarios such as logistics.

Among them, the real-person data collection glove can accurately capture the 21 joint degrees of freedom of the human hand, high-precision tactile information, and simultaneously record the visual data from the head-mounted and hand perspectives.

Compared with the data collection solutions of simulation data, teleoperation data collection, and UMI chosen by the industry, Lingchu Intelligence has always adopted the strategy of general dexterous operation. Psi-SynEngine directly collects the operation data of frontline workers in real work, rather than collecting data at high cost and low fidelity in a built data collection field. Its data acquisition cost is only 10% of that of the real-machine teleoperation solution.

In the future, Lingchu Intelligence plans to launch a portable crowdsourcing version, which is expected to further reduce this cost.

Meanwhile, the company also released a large-scale real-world multimodal dataset Psi-SynNet-v0 covering vision, language, touch, and actions. This dataset covers real operations in multiple industries, scenarios, objects, and skills, with a data volume reaching the same level as large language models. The data system has completed closed-loop verification in multiple industries where Lingchu Intelligence has implemented its solutions.

The core team of Lingchu Intelligence spans the age gradients of the 1970s, 1980s, 1990s, and 2000s, bringing together senior industry professionals and a large number of scientists.

Wang Qibin, the founder of Lingchu Intelligence, has nearly 20 years of successful operation experience in the mobile phone (BlackBerry), smart speaker (Sonos), and robotics fields (CloudMinds Technology, JD.com, etc.). He has achieved the industrial closed-loop from product definition, development, to market launch, and then to globalization from 0 to 1 to N multiple times.

Chen Yuanpei, the co-founder, is a post-2000s native from Guangdong who is obsessed with robots. As a visiting scholar at Stanford, he studied under Professors Karen Liu and Fei-Fei Li. He was the first in the world to use reinforcement learning to control the multi-skilled operations of both arms and hands simultaneously in the real world.

Lingchu Intelligence said that the data gloves have been successfully developed into hardware, and 100 sets of equipment in the Beijing area are being deployed. It is expected that large-scale data collection will start before the Spring Festival. Previously, the company has completed the verification of 10,000 hours of real glove data, and the goal for 2026 is to exceed 1 million hours. The company also plans to further promote the implementation of embodied intelligence in complex logistics environments this year and promote process-based delivery.

This article is from the WeChat official account "Robot Forward", written by Xu Lisi and edited by Mo Ying. It is published by 36Kr with authorization.