As technological maturity deepens and scene adaptability strengthens, the humanoid robot industry in Shanghai's Zhangjiang area is entering a pivotal phase of growing orders, with leading manufacturers rolling out new products and refining their commercial strategies.

"Our K2 humanoid robot commenced small-batch mass production on Aug 4, with an expected output of around 100 units this year, primarily targeting automotive logistics and educational sectors," revealed a spokesperson from Kepler Robotics.

The K2 robot, known for completing the world's first 8-hour endurance live broadcast at the 2025 World Artificial Intelligence Conference, has already secured orders from multiple automotive companies for logistics applications and five universities for research and education purposes.

Kepler's K2 robot. [Photo/Kepler]

Its "charge for one hour, operate for eight" capability perfectly matches the single-shift work schedule in automotive warehouses. "Some automotive clients reported that during testing, the K2 achieved 80 percent of manual material handling efficiency, prompting them to place trial orders," the spokesperson added.

Kepler is not alone in this surge. According to a recent report by IDC, a leading global IT market research firm, KEENON Robotics, the world's top provider of commercial service robots by volume, has deployed over 100,000 robots across 600 cities in more than 60 countries and regions.

"In the second quarter alone, our XMAN series humanoid robots secured new orders in 12 hotel and catering scenarios, primarily for greeting and food delivery roles," stated a market executive from KEENON. These orders represent "position-based service packages", where robots like the XMAN-F1 collaborate with cleaning and delivery robots to provide end-to-end services, from room tidying to item delivery.

KEENON's robot works as a chef. [Photo/KEENON Robotics]

Behind the increasing orders lies differentiated technological approaches tailored to specific scenarios.

"For rapid commercialization, humanoid robots should excel as 'position experts' rather than striving for all-around capabilities," explained a technical lead at KEENON, highlighting their "general + specialized" model. By dissecting catering scenarios into standardized roles like greeting, food delivery, and table clearing, each role corresponds to specific skill modules.

"For instance, the XMAN-F1's hand manipulation ability is tailored for food preparation and room tidying, ensuring stability while enabling quick adaptation to new roles through general-purpose AI models," the lead said, noting a 40 percent reduction in order fulfillment cycles for hotel and catering projects.