Hangzhou, in Zhejiang province, recently held a unique opening ceremony 鈥?marking the enrollment of its first batch of 30 "robot students." These machines come from fields such as industry, services, and security, and their "parents" include representatives of small and medium-sized enterprises as well as relevant government departments. The "Hangzhou Robot School" was jointly established by the Robotics Research Institute of Zhejiang University, the Zhejiang Institute of Quality Science, and leading companies in the industry. Here, robots will undergo systematic "vocational skills" training, learning application skills for various scenarios. After obtaining "skill certificates," they will move on to real jobs, completing the transformation from "tech toys" to "practical tools."

First, let's look at admission requirements. Cai Jiandong, vice president of the Robotics Research Institute of Zhejiang University, explained that before enrollment, each robot must pass an "entrance physical exam" 鈥?a comprehensive assessment of hardware performance, functional integrity, algorithm compatibility, and other aspects. As for the curriculum, the school follows the model of human vocational education and offers four specialized tracks: technical school, health school, art school, and sports school. Based on pre-enrollment test results and actual industry needs, training is divided by subject. The training covers five dimensions: ethics and safety, perception ability, motion performance, aesthetic expression, and practical scenario operation, aiming to comprehensively improve each robot's overall capabilities. Each "student" has a different training duration; the school tailors instruction to each individual, creating a personalized development plan for every robot. After training, before graduation, robots will undergo capability assessments by authoritative institutions and receive "specialized skill-level certificates" 鈥?a pass to enter specific industries, marking the leap from "bare metal" to "finished product."

Tong Fei, founder and CEO of Zhongcheng Yuanzhu Robot Technology Co., Ltd., sent his "child" to the school with high hopes: "The robot already has basic dance skills. I hope that through learning, its performances will become more human-like and its movements smoother."

Why establish a dedicated school for robots? Zhu Shiqiang, dean of the Robotics Research Institute of Zhejiang University and founder of the school, is blunt about the reason: many domestic startups possess robots that can run steadily and jump high, yet they fail on the path from "bare metal" to "product." As a result, although there is urgent demand in industrial manufacturing, domestic services, medical care, and other scenarios, robots still struggle to enter homes and daily life on a large scale. "The problem lies in the lack of a brain capable of developing and adapting to scenarios," Zhu said.

Training robots, the "brain" is key. Zhu Shiqiang's team independently developed the "Wuji Brain," a system that adds a "logical reasoning layer" to the traditional vision-language-action architecture, creating a complete cognitive loop. Zhu explained: "Simply put, in the past, robots relied on memory to handle scenarios. They could manage familiar situations but had no way to deal with unfamiliar ones. With the logical reasoning layer, robots can deduce physical laws, analyze task constraints, predict the consequences of actions, and then make decisions." From "memorizing answers" to "solving problems" 鈥?this is the skill robots learn at Hangzhou Robot School.

Among the first 30 robots enrolled, a guide robot named "Xiaoji" was a bare metal machine that could only run and jump half a month ago. After installing the "Wuji Brain," it learned skills such as guided tours, greeting interactions, equipment operation, and multilingual communication. When faced with random questions from unfamiliar visitors, it can understand intent, reason about the situation, plan routes autonomously, and complete guided tours accurately. "What they receive is not just skill certification, but a set of ability genes for continuous learning," Zhu said. Even more valuable, through brain empowerment and intelligent agent integration, robots can continue to learn and iterate after entering real-world scenarios.