Body, Size & Structure
• Dimensions and Height: ~120 cm.
• Weight: ~30 kg.
• Leg length (thigh + calf): ~57 cm (useful for judging stride, reach, stability).
• Arm reach: ~45 cm. 

 Standard configuration: 23 DoF — which includes 6 DoF per leg (so 12), 4 DoF per arm (so 8), 1 DoF for the waist, and 2 DoF for the head.

Extended configurations: with grippers or dexterous hands, the DoF can increase to 31 DoF (gripper version) or even 41 DoF (dexterous hands version) — significantly increasing manipulation potential

Joint mechanics: Uses full-force-controlled joints with dual-encoder modules, enabling mixed control modes — torque, velocity, and position control — which allow for precise, fluid, and dynamic motion.

Omnidirectional walking

forward, backward, sideways, turning

Dynamic motion & agility

e.g. smooth transitions like rising from prone (lying down) to standing, sports-style movements, and more complex maneuvers beyond simple walking.

Manipulation (when equipped)

with grippers or dexterous hands, capable of object handling, grasping, and potentially more advanced manipulation tasks

Robustness & resilience

the design is described as durable, shock-resistant, and capable of handling dynamic tasks such as in competitions or rough terrains.

Depth (RGB-D) camera

for 3D vision, environment mapping, object detection, navigation, etc.

9-axis IMU

for balance, orientation, motion sensing (gyro + accelerometer + magnetometer), crucial for stable walking, standing, recovery from disturbances.

Audio subsystem

microphone array + speaker — enabling speech recognition (ASR), text-to-speech (TTS), voice interaction, which helps in HRI (human-robot interaction), voice commands, interactive demos.

Optional modules / expandability

depending on configuration (e.g. grippers, dexterous hands), additional sensors/actuators may be included for manipulation tasks.

T1 is built around an open-source framework / SDK, giving developers access to both high-level abstractions (walking, behaviors, perception, AI) and low-level controls (joint commands, sensors).

Because it supports standard robotics middleware (ROS 2) and major simulators (Isaac Sim, MuJoCo, Webots), it’s friendly for labs, research groups, universities — facilitating testing, simulation-based development, and safe validation before deploying to real hardware.

With powerful onboard AI compute (200 TOPS), developers can run heavier perception models, vision-based navigation, object detection, speech recognition / generation, or even edge-LLM (some configurations support optional “miniCPM” edge LLM) — enabling advanced embodied-AI experiments.

It aims to deliver “real-world” humanoid capabilities (locomotion, manipulation, perception, AI) in a package that is still manageable (size, weight, modularity) — lowering the barrier for labs, universities, research groups, and competition teams.

By offering an open-source ecosystem (SDK + ROS 2 + simulation support) plus high onboard compute, T1 encourages rapid development, experimentation, and iteration of algorithms — from walking control to embodied AI — without requiring custom hardware design from scratch.

As a “champion-grade” humanoid: T1 has already been used in competitions, showing that it’s not just a research toy — it’s capable of dynamic, competitive scenarios.

Its modularity (hands, grippers, sensors) allows it to scale: from locomotion-only tasks to full manipulation + perception + interaction systems — making it a flexible platform for a wide range of robotics domains.