NHK Reports Growing Adoption and Development of Physical AI in Agriculture

On October 28, 2026, NHK reported on moves to introduce and develop physical AI in agriculture.

The report covered growing efforts to introduce and develop physical AI in agriculture. As information on specific equipment specifications and the scale of adoption is not available here, the reported developments need to be considered in terms of where the technical challenges lie in tasks involving crops. A related article from this newspaper covered a cucumber-harvesting robot involving Yaskawa Electric and JA, focusing on the challenge of judging when to harvest, with only “1 in 10” judgments correct Nikkei: Yaskawa Electric and JA’s cucumber-harvesting robot. This issue shows that agricultural robots must do more than move and grasp: they also need to assess the condition of their targets and determine whether to perform a task. However, it has not been confirmed that the specific development NHK reported on uses the same approach or faces the same challenge. Any connection with Yaskawa Electric’s factory-oriented technology should also be considered in terms of both commonalities and differences. This newspaper has reported that the company’s “MOTOMAN-PA3050HB” is a seven-axis robot with a maximum reach of 3050mm and a payload of 15kg Yaskawa Electric launches seven-axis robot “MOTOMAN-PA3050HB”. The CPU-14U added to the MPX1010 machine controller supports control of up to 256 axes Yaskawa Electric adds CPU-14U to MPX1000 series, supporting control of up to 256 axes. These are specific examples of industrial robots and control systems; they do not demonstrate that crop recognition, decision-making and harvesting in fields have already been achieved. To assess agricultural applications, crop-condition recognition, harvest decisions and the reproducibility of actual work need to be examined separately. The industry’s structural challenge lies not only in the robot itself, but also in combining crop-recognition systems with harvesting movements and adapting to changing field conditions. A track record in controlling factory equipment does not automatically guarantee applicability to differences in crop conditions or field environments. What remains to be confirmed is the specific technology, scale of deployment, work targets and operational status at worksites mentioned in the report. These details cannot be confirmed from the information available to this newspaper. See NHK’s original report for details.