
The Unitree G1 Base is a small-form humanoid aimed at education, demonstrations, and light R&D. It focuses on reliable bipedal mobility, basic manipulation, and an approachable workflow rather than open-ended research features. If you need a concise spec overview and current configurations, see the Unitree G1 Base product page.
In standing posture the G1 measures about 1270 × 450 × 200 mm and folds to 690 × 450 × 300 mm for transport. With the battery installed, mass is around 35 kg. The Base model provides 23 total degrees of freedom (DoF): 6 per leg, 5 per arm, and 1 at the waist. This layout gives stable walking, turning, and simple upper-body gestures without the complexity of finger-level actuation.
Each joint uses a high-response PMSM actuator with dual encoders and crossed-roller bearings for rigidity. The knee joint delivers up to 90 N·m peak torque on the Base, supporting step-ups, controlled crouches, and recovery motions. The arm load capacity is rated around 2 kg, which is suitable for lightweight props, tools, and teaching fixtures.
For environment awareness the G1 integrates a depth camera and 3D LiDAR, forming a straightforward perception stack for pathing and obstacle checks. Audio hardware includes a four-mic array and onboard speaker for feedback cues. Control runs on an 8-core high-performance CPU, and the platform supports Wi-Fi 6 and Bluetooth 5.2 for connectivity plus OTA firmware updates.
A quick-release 13-series Li-ion battery (≈9000 mAh) powers the robot for about two hours under light-to-moderate workloads. Charging is via a 54 V / 5 A supply. For extended sessions, teams typically rotate multiple packs rather than pushing a single battery to empty.
Operation is designed to be straightforward: manual controller for basic motions, preset actions for demos, and step-by-step routines students can iterate on. The Base is ideal for introductory biped labs, campus outreach, and public demonstrations where reliability and repeatability matter more than deep API access. Advanced secondary development, higher joint counts (up to 43 DoF), and dexterous three-finger hands are available on EDU configurations and as options—useful to plan for if you anticipate research into manipulation.
Humanoid robots store significant energy in their joints. Keep clear safety perimeters, use trained handlers, and start on flat, uncluttered floors before moving to ramps or steps. Build a pre-run checklist (battery state, joint temperature, comms link, emergency stop) and log firmware versions so multiple operators can maintain consistent behavior.