Robots Atlas>ROBOTS ATLAS
Robotics & Hardware

Galbot Teases ET1 Bipedal Humanoid Ahead of WRC 2026

Lady Robot23 August 2026 · 3 min read
Galbot Teases ET1 Bipedal Humanoid Ahead of WRC 2026

Chinese startup Galbot teased its first bipedal humanoid, ET1, on August 18, 2026. For now it is only a CGI render, not footage of real hardware. The formal debut is planned for the World Robot Conference 2026 in Beijing and marks a clear pivot from wheeled platforms toward walking humanoids.

Key takeaways

  • The teaser is a CGI render (the robot performs a tennis swing), not footage of real hardware
  • Debut at WRC 2026 in Beijing, August 19-23, 2026, booth C104
  • Galbot shipped about 900 units in the first half of 2026 (its wheeled G1 platform)
  • The company is backed by CATL and is exploring a Hong Kong IPO valued at up to USD 4 billion
  • Specs undisclosed: degrees of freedom (DOF), actuators, sensors, weight and payload
ET1 has so far been shown only as a CGI render — no live demo and no disclosed technical specs.
~900units of the wheeled G1 platform shipped in the first half of 2026Humanoids Daily

A teaser, not a launch

The clip is a cinematic render showing a white humanoid with a smooth black visor instead of a face, executing a tennis swing. There is no live demo and no footage of a physical machine. Galbot is keeping the technical details under wraps — no degrees of freedom, actuator architecture, sensor suite, weight or payload. At this stage ET1 should be read as a statement of direction, not a finished product.

From wheeled platforms to bipeds

Galbot's current product is the G1, a wheeled dual-arm platform for retail and logistics tasks. In the first half of 2026 the company shipped about 900 units and ranks among China's leading vendors, alongside AGIBOT and Unitree. ET1 is therefore an architectural pivot: a move from a stable wheeled base to harder bipedal locomotion, driven in part by pressure from rivals releasing their own humanoids.

Tennis as the bar

The choice of a tennis swing is not accidental. With researchers from Tsinghua and Peking University, Galbot developed the LATENT framework — hierarchical reinforcement learning for dynamic tennis skills. It was first tested on a third-party platform, the Unitree G1 humanoid. The tennis benchmark itself demands sprint speeds above 6 m/s, millisecond-level visual feedback loops and dynamic full-body balance — exactly the capabilities wheeled platforms lack.

Why it matters

A CGI teaser shows ambition, not proof of function. For Galbot it signals a bid to play in the walking-humanoid league, where Unitree and AGIBOT already have physical machines in production. The real test will be a first live demo rather than a render — only that will show whether the company can turn locomotion research into working hardware.

What's next?

  • WRC 2026 in Beijing (August 19-23, booth C104) is the nearest chance for a possible live ET1 demo
  • Galbot is exploring a Hong Kong IPO valued at up to USD 4 billion, pointing to plans for rapid scaling

Sources

Share this article