en.Wedoany.com Reported - Chinese robotics startup MindOn has successfully demonstrated its artificial intelligence system, Mind-0, which can simultaneously control multiple types of robots in a single logistics task. In the demonstration, humanoid robots and fixed robotic arms collaborated using a unified AI model, marking the company's progress toward achieving robot intelligence independent of specific hardware.

MindOn's core innovation lies in its training data source. Instead of collecting data by manually controlling robots, the company uses human motion capture data to train its models. This human-centric training method is believed to better preserve humans' natural problem-solving abilities. To convert human actions into robot actions, MindOn has developed a cross-morphology data pipeline that transforms human movements into representations usable by different robotic systems, supported by a full-body motion model trained on a large motion capture dataset, enabling robots to balance, coordinate, and execute tasks according to their physical constraints.

To address the common "sim-to-real" challenge in robotics, the team introduced a real-world execution compensation model. This system uses a small amount of real-world deployment data to correct performance discrepancies between simulation and physical hardware. MindOn claims that this technology achieves sub-centimeter operational precision on the Unitree G1 humanoid platform. Additionally, its system includes a hierarchical reasoning module that compensates for mechanical execution delays by continuously monitoring feedback from underlying controllers and adjusting instruction timing in real time, ensuring synchronization between planning and execution.

The company recently conducted a logistics demonstration using a mixed formation of Unitree G1 humanoid robots and fixed dual-arm robotic arms. The robots collaborated to complete the full workflow of picking, transporting, packaging, and sealing boxes. Despite different physical configurations, all robots operated based on the same AI model, proving Mind-0's potential as a universal solution for multi-robot platforms.

Founded in Shenzhen in 2025, MindOne Robotics positions its technology as a scalable alternative to traditional teleoperation-dependent solutions. The company plans to expand its human-centric dataset and extend its applications to various robot forms, including mobile dual-arm systems. Its ultimate goal is to create a universal robot intelligence that can run on multiple hardware platforms.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com