Toborlife AI announced a commercial teleoperation system for the Unitree G1 Edu humanoid that pairs its Tobor Harness control system with a PICO VR headset operator interface. The system lets an operator move naturally and have the G1 mirror head, limbs and torso motion in real time. Where present, G1's optional dexterous robotic hands can be teleoperated to perform finger-level grasps and manipulation.
- Real-time full-body teleoperation: The operator's movements translate to joint-level control on the G1 without intermediate scripting.
- On-premises data sovereignty: Recorded demonstrations stay on customer hardware in the United States. Toborlife AI does not access or transmit customer data unless the customer requests support.
- US-based support and onboarding: Deployments are backed by Toborlife AI engineers located in the US who provide configuration guidance, onboarding and hands-on troubleshooting.
Researchers and developers in the US have been acquiring capable humanoid hardware like the G1 but often lack a domestic software layer tailored for data collection and model training. The first months with new humanoid platforms frequently go to integrating disparate open-source libraries and building infrastructure before any experiments start. Toborlife AI positions its teleoperation stack as that missing layer, allowing teams to begin recording demonstrations and training skills earlier.
An operator wears a PICO VR headset and performs tasks. The G1 mirrors the operator's movements in real time. The session is recorded and exported in formats suitable for imitation learning or other training pipelines. Toborlife AI provides setup support and can assist if customers request troubleshooting that requires access to datasets.
Teleops is offered as an end-to-end data collection and training stack tailored to domain-specific applications. Toborlife AI frames the system as a foundation that lets teams focus on building skills for specific use cases rather than on infrastructure work. The announcement comes as more Chinese-built humanoid platforms, including Unitree's G1, appear in US labs and researchers seek domestic software and service options.
Practical next steps for interested teams
- Evaluate hardware compatibility: Confirm Unitree G1 Edu configuration and whether optional dexterous hands are included.
- Plan for on-premises storage: Ensure local hardware meets capacity and security needs for motion dataset storage.
- Schedule onboarding: Coordinate with Toborlife AI's US engineering team for setup and training.
- Define target tasks: Identify domain-specific demonstrations to record so early data collection yields immediately useful training sets.
Toborlife AI's teleoperation release aims to shorten the time between acquiring a Unitree G1 Edu and producing high-quality demonstration datasets for training. The package combines real-time control, finger-level capture, on-premises data retention, and US-based support to address immediate practical needs of research labs and developers.