The robot is equipped with a 275 TOPS computing unit and a multi-sensor fusion system including LiDAR, a depth camera and a wide-angle camera. It can patrol preset routes, move flexibly between multiple guard posts, and cover blind spots left by fixed surveillance.

　　"Marathon security is concentrated in time, with dense crowds and highly repetitive tasks, which makes it a good scenario for robots to take on first," Meng Yuan, a representative of DEEP Robotics, told the Global Times on Sunday.

　　"This deployment is also an exploration of a standardized application model for humanoid robots in traffic management at large marathon events."

　　The robot uses multi-sensor environmental perception and a high-precision motion control system to detect traffic violations such as pedestrians crossing the road and vehicles cutting in, in real time. While carrying out standardized traffic control gestures, it also automatically broadcasts traffic safety advisories, helping police officers manage pedestrian flows more efficiently, according to the company.

　　The deployment is a notable example of how embodied artificial intelligence is being integrated into urban public services. It also comes as humanoid robots move beyond laboratories and into real-world use cases.

　　Chinese robotics company Unitree Robotics unveiled its new H2-X humanoid robot Kung Fu Legend Guardian in October. According to the company, the robot can potentially be applied in security and rescue scenarios, such as shielding a person from an attack, lifting the front of a car with its bare hands, and responding to a simulated traffic accident.

　　In April, traffic police embodied intelligence robots were also deployed at the Beijing Yizhuang Half Marathon, directing participants on the course.