←News detail

Shanghai ObserverUpdated Nov 3 13:08

Wuzhen "Light of the Internet" Expo: Robots from badminton sparring partners to massage therapy

AI summary

The 2026 World Internet Conference "Light of the Internet" Expo opened in Wuzhen, where 671 Chinese and foreign enterprises and institutions from 54 countries and regions are showcasing more than 1,000 new technologies and achievements, with robots shifting from fixed-motion performances to real-world scenario services. Dongyi Technology's PHYBOT C series bipedal humanoid robot can rally continuously with a person for 30 to 40 shots in badminton; the Yiqiwilai Zhizhen AI massage robot R2s carries a terminal price of 128,000 yuan, with full-year sales in 2026 expected to reach more than 1,000 units. Jingang Technology's six-legged robot "Tiaotiao" is equipped with 22 self-developed miniaturized tactile servos and plans to open up its AI training platform.

Body · AI translation

Robots accompany people playing badminton, provide massage therapy, and transform into little household helpers to tidy up rooms... Walking into the exhibition hall, one obvious impression is that robots are no longer just "performers" demonstrating fixed movements, but have entered real scenarios, interacting with people and providing services, exploring more possibilities from "being able to move" to "being able to work." The 2026 World Internet Conference "Light of the Internet" Expo opened on the 2nd in Wuzhen, Zhejiang. This year's expo runs for four days, gathering 671 Chinese and foreign enterprises and institutions from 54 countries and regions, showcasing and releasing more than 1,000 new technologies and new achievements. At the Dongyi Technology booth, a "human-robot badminton battle" is staged (photo by our reporter Guo Fanxi). Humanoid robot challenges badminton rallies. Spectators stand in the designated area to serve; as soon as the shuttlecock flies out, the 1.35-meter-tall bipedal humanoid robot quickly adjusts its posture, either dashing forward in small steps, or turning sideways to swing the racket, or leaning back to catch the shuttlecock. Whether forehand or backhand, the robot can return the shuttlecock fairly effectively. After hitting the return, it quickly readjusts its body's center of gravity and re-enters the ready state, waiting for the next incoming shuttlecock, quite resembling a badminton ace. In the embodied intelligence exhibition area, this exhilarating human-robot badminton match attracted many spectators to stop and watch. "Under suitable court conditions, the robot can already rally continuously with a person for thirty to forty exchanges," introduced Wang Mengdi, brand director of Dongyi Technology. This time the company brought the PHYBOT C series bipedal humanoid robot, demonstrating two highly dynamic sports scenarios: autonomously rallying with people in badminton and table tennis. The seemingly simple act of returning the shuttlecock tests the robot's comprehensive motion control capabilities. The shuttlecock flies fast and its landing points vary greatly; the robot must complete observation, judgment, movement and hitting within an extremely short time, while coordinating its legs, torso and arms to avoid losing balance during rapid movement. To this end, Dongyi Technology developed a training process based on multi-stage reinforcement learning, allowing the robot to gradually transition from footwork movement and racket-swinging motions to actual rallying, rather than relying on complex pre-programming or simply imitating human movements. The team also used trajectory prediction and instant response systems to help the robot judge the direction of the incoming shuttlecock, adjust its position in advance, and complete the return amid dynamic changes. In Wang Mengdi's view, the value of the badminton scenario does not lie in letting the robot defeat humans, but in verifying whether autonomous decision-making and whole-body motion control capabilities can work together in a real environment. The team chose badminton both because it is a sport with a broad grassroots following and because they hope to respond to the daily-life need that "it's hard to find a badminton partner." Previously, the team held a one-month pop-up event in a Guangzhou shopping mall, attracting nearly 10,000 citizens to participate in human-robot rallies, and implemented interactive experience scenarios in a technology park. In the future, the PHYBOT C series also hopes to expand into applications such as sports teaching assistance and sports companionship. Robotic arms replicate the techniques of professional massage "masters." AI massage robots are likewise busy as can be. In the Yiqi Weilai exhibition area, rows of robotic arms are providing massage therapy services to visitors. The ends of the robotic arms move along the curves of the human body, kneading and pressing different parts of the back through techniques such as finger stroking, finger kneading and finger pushing. Visitors can also adjust the massage intensity and speed through voice commands. A programmer who has long worked hunched over a desk exclaimed after trying it: "The effect is no worse than a veteran master; I strongly demand that every big company put one in." To make the robotic arm press accurately and steadily, it must first understand the shape of the human body. Wang Dongliang, co-founder of Yiqi Weilai, introduced that the team relies on 3D vision models to scan and model the human body, identifying people's height, build and body curves, so that the robotic arm can fit the human contours in real time during the massage. In terms of control precision, the team raised the robotic arm's control frequency to 1,000 Hz, that is, a control adjustment every millisecond, to improve motion continuity and reduce the feeling of stuttering. More importantly, the robot does not simply execute movements according to preset trajectories. The team used motion capture equipment to collect the real techniques of professional masseurs, recording force, angle, motion trajectories and even subtle changes, and then converted them into the robot's execution capabilities through imitation learning. "However the master moves, however they trace on the human body, we collect these movements exactly as they are," Wang Dongliang said. From wellness centers to elderly care institutions, fitness venues and rehabilitation institutions, this type of product is trying to enter more health service scenarios. Currently, Yiqi Weilai's products are mainly aimed at enterprise customers, and some products have also entered households. Wang Dongliang introduced that, taking the Zhizhen AI massage robot R2s as an example, its terminal price is 128,000 yuan, and the company expects full-year sales in 2026 to reach more than 1,000 units. Small robots target family life. Swaying its body to a dynamic rhythm, its six legs hitting the beats precisely, with light and coherent movements—a crab named "Tiaotiao" that can do sidestep shuffles has become the "scene-stealer" of the exhibition area. "Unlike large humanoid robots, we hope this small six-legged robot can enter life scenarios closer to ordinary people with its compact size and agile movement capabilities," introduced Li Jingyu, brand publicity director of Jingang Technology. "Tiaotiao" is equipped with 22 self-developed miniaturized tactile servos across its body; the servos are as small as a thumb, and each integrates two pattern-type magnetic encoders: one senses the position of the motor output end, and the other senses the position of the joint end. At the same time, its unique crab-claw shape can also help grip objects and hand things over. This design allows the servo to evolve from "a muscle that executes commands" into "a muscle with its own neural reflexes." Li Jingyu explained that traditional solutions rely on a single encoder, which can only know how many turns the motor has made but cannot accurately perceive the actual angle the joint has reached. The dual-encoder architecture, by contrast, achieves full-chain position and force perception, allowing each servo to "feel" in real time whether the joint has encountered resistance, whether it has stepped on uneven ground, and whether it has gripped an object. As for the future of "Tiaotiao," the company's vision goes beyond dancing and display. Li Jingyu said the team hopes to continue challenging complex movements such as jumping, while also letting small robots gradually adapt to real home environments, for example jumping onto a bed to fold quilts or onto a sofa to fold clothes. The company also plans to open an AI training platform so that children can learn hands-on how to use AI to control robots and experience the combination of robot motion control and artificial intelligence. Starting from the six-legged form, Jingang Technology has also planned small robots in different forms such as four-legged and bipedal, which are expected in the future to target scenarios such as family companionship, children's programming education and scientific research and development, and to explore applications such as pipeline inspection.

Source

Shanghai Observer