Unitree Robotics has officially introduced the G1-D, its first wheeled humanoid robot designed for advanced data collection and model training. The robot serves as the centerpiece of a full-stack solution that combines high-performance humanoid hardware with a scalable software platform for data acquisition, AI training, and real-world deployment.
Specifications
The G1-D is available in two versions: Standard and Flagship. Both models stand between 1260 mm and 1680 mm tall and weigh up to 80 kg. The Standard version is stationary, while the Flagship version features a mobile base that supports differential drive and can move at speeds of up to 1.5 meters per second. The mobile base includes sensors such as LiDAR, depth cameras, and collision detection units.
Unitree has equipped the G1-D with 17 degrees of freedom on the Standard model and 19 on the Flagship model, excluding the end effectors. Each arm has seven degrees of freedom and supports a payload of up to 3 kg. The robot’s waist joint allows motion along the Z-axis up to ±155 degrees and along the Y-axis between -2.5 and +135 degrees. The platform supports a vertical working range of 0 to 2 meters.
The robot features a high-definition binocular camera on its head and an additional HD camera on each wrist. It supports various end effectors, including a two-finger gripper, three-finger dexterous hands with or without tactile sensors, and a five-finger dexterous hand. Unitree has powered the G1-D with an Nvidia Jetson Orin NX module that delivers up to 100 TOPS of AI performance. The Flagship version offers a battery life of up to six hours.
The G1-D is designed for a wide range of real-world applications. It can perform repetitive manual tasks in industrial and service environments, assist in warehouse automation, collect and annotate data for AI training, and even interact with retail or consumer settings using its onboard vision and mobility. The robot’s dexterous hands and mobile base allow it to manipulate objects, conduct inspections, or serve as a data-gathering agent in complex environments.
Alongside the hardware, Unitree has launched a comprehensive software stack that manages the entire AI workflow. This includes data acquisition tools, task management, annotation systems, and real-time tracking. The platform supports distributed training and integrates with popular open-source frameworks such as GROOT and PI. It also includes a simulation environment for model evaluation and supports seamless export and deployment of trained models.
In related news, Alibaba has recently unveiled its first self-developed AI glasses featuring swappable batteries, a POV camera, and the Qwen assistant. Meanwhile, an LSE study predicts that by 2028, Gen Alpha will rely on voice rather than keyboards at work.
For more daily updates, please visit our News Section.
Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories! 💡
(Source)
Comments