· Research & technology· 日本語で読む
TechXplore reports on Shenzhen robot training linked to human data
TechXplore published an article about robots in Shenzhen. The article focuses on training robots using human data. Related articles from this newspaper cover both robot models and physical robots.
What happened
TechXplore published an article about robots in Shenzhen on November 18, 2026.
Analysis
The key issue is not a new robot model or a mass-production plan, but how to connect models for robots with actions performed by physical robots. The article focuses on robot training using human data, but the text provided does not include figures that would allow comparisons of data scale, collection conditions or post-training performance. Therefore, the effectiveness of the training method and its stage of practical use cannot be determined from this article alone.
A related article from this newspaper, humanoid.guide publishes “Welcome, Spirit v1.5!”, also focuses on a model for robots, but the summary provided does not confirm technical details or implementation status. By contrast, X Square Robot’s mobile dual-arm robot “Quanta X1 Pro” named a CES 2027 Honoree describes the robot’s range of tasks, a design that does not require modifying the environment, and the foundation model “WALL-B.” These are separate reports, and there is no basis for linking the training method in the current article to Spirit v1.5 or WALL-B. Models and physical robots should not be assessed on the same terms; training methods, deployment and performance in real-world environments need to be examined separately.
Structurally, the key question is how much of the process—from human motion data as input, through model processing, to execution by a robot—is confirmed by the article. Without details on the tasks and range of environments covered by data collection, data-use conditions, model-update procedures, and success rates and safety on physical robots, it is not possible to assess how data acquisition translates into practical performance. In particular, whether data collected in real-world environments can be applied to other robot platforms or tasks, and whether execution results are fed back into retraining, are important questions for determining whether robot capabilities can be improved continuously. These points cannot be confirmed from the text provided; refer to the original source for details.
Sources
Translated from the Japanese original: /news/techxplore-shinsen-no-robotto-kunren-wo-ningen-no-deeta-to-m-9344a8. The Japanese version is authoritative if they differ.