en.Wedoany.com Reported - The X-Trainer industrial-grade data acquisition device launched by China's Dobot is becoming a key hardware foundation connecting embodied intelligence teaching, scientific research, and industry. This industrial-grade data acquisition platform addresses the data acquisition quality gap, model research gap, and application training gap by providing 25Hz synchronous teleoperation acquisition and industrial-grade dual-arm collaboration precision, streamlining the entire process from data benchmark construction and model training to algorithm validation.
The X-Trainer has been adopted by dozens of universities and research institutions worldwide. In China, Shanghai Jiao Tong University uses it to evaluate VLA models and iterative algorithms; the Chinese University of Hong Kong studies dual-arm robots from language commands to precision operations; the Hong Kong University of Science and Technology trains dual-arm collaborative robots to autonomously learn manipulation strategies based on it; Xidian University leverages digital twin technology to build scenarios such as logistics sorting; and Shandong University uses dual-arm collaboration to solve the challenge of barcode scanning in logistics transportation. In Europe, the Technical University of Munich in Germany completed environment setup and data collection to train humanoid robot models within a single day; the University of Koblenz in Germany developed industrial automation solutions; and the University of Bristol in the UK uses it for education, research, and industrial pilots. In the Americas, Northwestern University in the United States conducts dual-arm and multimodal perception research and has published multiple papers, while Syracuse University in the US has deployed it in its Smart Manufacturing Research Center.
In terms of scientific research output, over 20 global research institutions have conducted embodied intelligence research based on the X-Trainer, resulting in more than 10 published papers. Shanghai Jiao Tong University and Ant Group proposed the GM-100 benchmark and open-sourced over 13,000 real teleoperation trajectory data entries; the Mohamed bin Zayed University of Artificial Intelligence and others proposed the A0 hierarchical diffusion model; Northwestern University proposed the DreamTacVLA framework; and Tsinghua Shenzhen International Graduate School and Tencent Robotics X Lab proposed the Hi-ORS post-training paradigm.
In terms of application validation, the X-Trainer completed a snowball throwing project under conditions of minus 20 degrees Celsius, and 192 teams participated in simulation and real-machine competitions at the Dobot Embodied Intelligence Challenge. On the industrial side, this industrial-grade data acquisition platform has completed autonomous packaging in a cosmetics factory, switched between bottle and pouch packaging specifications on a laundry detergent brand production line, and completed the picking and sorting of multi-specification parts in a metal processing factory. Based on the VLA model, the X-Trainer has also completed complex sequential tasks from rice retrieval and ingredient preparation to cooking. Dobot stated that the X-Trainer has closed the loop across teaching, scientific research, and industry, driving embodied intelligence technology from the laboratory to real-world scenario applications.
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
