The Booster T1, developed by Booster Robotics (Beijing Jiashu Jinhua Technology Co., Ltd.) and first unveiled at the 2024 World Robot Conference, is a lightweight, high‑performance, open‑source humanoid robot built for developers, researchers, and competitive robotics teams . Designed with the core philosophy of being “lightweight, agile, and exceptionally durable,” the T1 excels in complex motion tasks while maintaining robustness against falls and real‑world disturbances . It is not a demonstration prototype — it is a battle‑tested platform that has enabled the Tsinghua Hephaestus team to win the RoboCup 2025 AdultSize category championship in Brazil, the first time a Chinese team secured the top honor in that division .
Key Features
Open‑Source Architecture with Complete API & ROS 2 Compatibility
The Booster T1 is built around an open‑source framework and SDK, giving developers access to both high‑level abstractions (walking, behaviors, perception, AI) and low‑level controls (joint commands, sensors) . It offers full ROS 2 compatibility and support for major simulation environments including Isaac Sim, MuJoCo, Webots, making it a true development platform for embodied AI research, algorithm testing, and simulation‑to‑real deployment . Advanced multimodal interactions are supported, including optional edge LLM integration, TTS (Text‑to‑Speech), ASR (Automatic Speech Recognition), and recognition algorithms such as YOLO .
Compact, Lightweight, and Durable Form Factor
Standing approximately 118 cm tall (118 × 47 × 23 cm) and weighing only 30 kg, the Booster T1 is designed to be handled by a single person — unlike larger humanoid robots that require several people to move them . The high‑strength metal and engineering plastic chassis ensures durability against wear, tear, and repeated falls. The robot features a leg length of 57 cm and an arm length of 45 cm, with a battery life of approximately 2 hours of continuous walking and 4 hours in standing mode .
Configurable Degrees of Freedom – Three Versions for Different Application Needs
The Booster T1 is available in three configurations to match different research and application goals:
| Configuration | Total DOF | Per Arm DOF | End Effector | Best For |
|---|---|---|---|---|
| Standard | 23 | 4 (per arm, extensible) | None | Locomotion research, perception studies, human‑robot interaction |
| With Gripper | 31 | 7 (per arm) | Electric grippers (1 DOF each) | Pick‑and‑place manipulation, object handling, utilitarian grasping |
| With Dexterous Hand | 41 | 7 (per arm) | 6‑DOF dexterous hands (5 fingers each) | Fine manipulation, advanced dexterity tasks, complex grasping |
The Standard version is suitable when the priority is locomotion, perception, and interaction without advanced manipulation goals. The Gripper version targets utilitarian grasping and pick‑and‑place applications. The Dexterous Hand version is designed for finer manipulation that requires more tuning and calibration for scenarios and control. All three versions share the same DOF breakdown: 6 DOF per leg, 1 waist DOF, and 2 head DOFs .
Powerful Onboard AI Computing – NVIDIA Jetson AGX Orin
The Booster T1 is powered by an NVIDIA Jetson AGX Orin platform paired with an Intel i7 1370p CPU, delivering up to 200 TOPS of AI compute . This high‑performance computing enables onboard perception, deep‑learning inference, and real‑time control for running heavier perception models, vision‑based navigation, object detection, speech recognition and generation, and even edge LLM deployment (some configurations support optional “miniCPM” edge LLM) — enabling advanced embodied AI experiments .
Advanced Motion and Locomotion Capabilities
The T1‘s motion system is designed for agile, robust bipedal locomotion. The robot achieves omnidirectional walking (forward, backward, sideways, and complex rotations), resistance to external disturbances on uneven surfaces, and the ability to absorb moderate impacts without losing balance . It supports predefined programmable actions such as waving, dancing, push‑ups, kung fu routines, and kicking — and can go from lying down to standing up in record time . The robot features a maximum peak joint torque of 130 N·m at the knees, with dual‑encoder actuator design for precise control and force feedback .
All‑Round Multimodal Perception
-
Depth camera – RGB‑D visual perception for obstacle detection, object recognition, and spatial mapping
-
9‑axis IMU – Real‑time orientation, acceleration, and stability monitoring for balance control
-
6‑microphone circular array – 360° far‑field voice pickup for speech interaction
-
Integrated speaker – Natural‑sounding voice output for dialogue and feedback
Robust Joint Actuators with Wide Motion Ranges
The Booster T1‘s joint actuators are engineered for wide motion ranges, enabling human‑like flexibility across all major joints:
| Joint | Motion Range |
|---|---|
| Waist | ±58° |
| Hip | Pitch: ±118°, Roll: -21° ~ +88°, Yaw: ±58° |
| Knee | 0° ~ +123° |
| Ankle | Pitch: -50° ~ +20°, Roll: ±25° |
Software Ecosystem and Development Tools
The Booster T1 comes with a complete development ecosystem to accelerate research and application building:
-
Comprehensive API – High‑level motion APIs, low‑level hardware APIs, and status feedback interfaces
-
Simulation support – Full compatibility with Isaac Sim, MuJoCo, Webots
-
ROS 2 compatibility – Quick operation and monitoring of the robot within ROS ecosystems
-
Mobile app connectivity – Real‑time control and feedback via Bluetooth connection
-
Booster Gym – Open‑source reinforcement learning framework for locomotion training
-
Firmware upgrades – OTA update support for continuous improvement
Dexterous Hands (Optional) – RH56DFX
For applications requiring advanced manipulation, the T1 can be equipped with RH56DFX dexterous hands, featuring:
-
6 degrees of freedom per hand with 12 total joints
-
Active fingertip gripping force ≥ 3 kg
-
17+ tactile sensors per hand, refresh rate ≥ 30 Hz
-
Integrated force sensors for precise force‑controlled grasping
-
ROS compatibility with ready‑to‑use ROS plugins
-
Weight: 800 g per hand
Gripper (Optional) – EG2-4C2
The EG2-4C electric gripper is available for utilitarian grasping applications:
-
Communication interface: RS485
-
Total stroke (both sides): 70 mm
-
Weight: 231 g
-
Grip force: 0 to 20 N, accuracy ±1 N
-
Maximum speed: 70 mm/s, full stroke closing time 1.3 s
-
Protection level: IP40, operating temperature 0 to 40 °C
-
Automatic locking system in case of power cut
Key Application Scenarios
| Application | Description |
|---|---|
| Academic Research & Education | Locomotion algorithms, reinforcement learning, computer vision, whole‑body control, embodied AI research — used by Purdue, UC Berkeley, Carnegie Mellon, Tsinghua |
| Competitive Robotics | The championship‑winning platform of the 2025 RoboCup AdultSize Soccer category, trusted by 50+ global robotics teams and research institutes |
| Industrial Research & Development | Testing and developing innovative robotic solutions for warehouse automation, logistics, and manufacturing applications |
| Human‑Robot Interaction | Natural dialogue, gesture recognition, proactive engagement, and multimodal interaction experiments |
| Educational Demonstrations | Interactive tool to introduce students to advanced robotics concepts, programming, and real‑world humanoid deployment |
| Entertainment & Events | Pre‑programmed dance performances, sports‑themed demonstrations (kicking, martial arts), brand ambassador roles |
| Robotics Competitions | RoboCup Humanoid League and other bipedal robotics competitions requiring agile locomotion and team coordination |
Proven Commercial Deployment and Global Reach
The Booster T1 has demonstrated real‑world deployment success across multiple continents:
-
700+ units shipped globally as of early 2026
-
200+ customers across 20+ countries, with over 50% of orders from international markets
-
50+ top global robotics teams and research institutes have selected the T1 as their preferred platform
-
100+ research institutions served globally
-
Complete mass production delivery with mature toolchain supporting complex development scenarios
-
Championship pedigree: RoboCup 2025 AdultSize Champion (Tsinghua Hephaestus), IEEE-RAS Humanoids 2024 Champion
Why Choose Booster T1?
-
True open‑source platform – Full API, ROS 2 compatibility, and comprehensive SDK access — not a locked‑down system
-
Battle‑tested reliability – Championship‑proven locomotion system with fall recovery, push‑ups, kicks, and kung fu — validated in RoboCup‘s most demanding environment
-
Affordable for research – Market‑competitive pricing makes advanced humanoid robotics accessible to academic labs, research institutions, and development teams
-
Lightweight and single‑person operation – Only 30 kg, can be handled, transported, and deployed by one person
-
Production‑ready, not a prototype – Mature mass‑production capability with stable supply chain and global delivery
-
Modular and configurable – Three end‑effector options (none, grippers, dexterous hands) allow the same chassis to serve multiple research directions
-
Global developer community – Trusted by 50+ robotics teams worldwide, with active knowledge sharing and community support
-
Proven market validation – Over 700 units shipped globally with 50%+ international sales — real demand, not speculation
There are no reviews yet.