
Full size
Height 176 cm, weight 70 kg, 52 independent axes of motion, humanoid gait, walking speed 2 m/s.

UBTECH Walker S2
Dahuan PGC-140-50 · 40–140 N · hod 50 mm · 1,5 kg · 48 V DC · EtherCATGripping force is adjustable; fits the arm end flange.pair
Intel RealSense D405C · 7–50 cm · 1280×720 @ 30 FPS · H84° × V58° · 42×42×22 mm · USBGives the robot and the operator an extra viewpoint, improving grasp accuracy and success rate.pair
Pico 4 UltraSupports upper-body teleoperation.set

About this model
Walker S2 is a humanoid robot developed by UBTECH for smart services and assistance in the home. Its characteristics are advanced bipedal walking, dexterous hand movements and the ability to communicate with people using voice and face recognition.
With improved balance and mobility, Walker S2 can carry out complex tasks such as delivering objects, opening doors and even playing chess.



Height 176 cm, weight 70 kg, 52 independent axes of motion, humanoid gait, walking speed 2 m/s.

Pitch angle +90° to −35°, rotation angle ±162°.

LLM (large language model) support, voice interaction, 4-inch round display, face display, four-microphone array, two speakers.

11 independent axes of motion, six-sensor tactile array, single-hand grip 7.5 kg, finger grip 1 kg.

Hybrid rigid-flexible structure, heterogeneous and composite materials, aerospace-grade aluminium alloy, 3D-printed primary frame, 3D-woven high-elasticity fibre material.


A dual battery hot-swap design lets Walker S2 switch between dual-battery and single-battery operating modes. By integrating real-time battery management with battery swap stations, Walker S2 independently chooses whether to swap or charge the battery according to task priority, achieving dynamic power management.


With a human-like whole-body dynamic balancing algorithm, Walker S2 reaches a 170° lean and carries a 15 kg load across a 0–1.8 m working range.

A complete RGB binocular stereo vision solution is built into the head of Walker S2. The in-house passive binocular vision system uses deep-learning depth-estimation algorithms to produce accurate, left-aligned dense depth maps in real time. This lets Walker S2 achieve stereoscopic perception similar to the human eye.



