A humanoid platform makes sense for your own research when the interfaces and equipment of the version on offer suit your planned experiments — not simply because it can demonstrate movements. According to RBTX and Unitree, the Unitree G1 EDU offers documented development access. For grasping experiments, a Booster T1 variant with a gripper or hand may also be suitable. Before buying, confirm what’s included, which safety measures are needed for your specific lab set-up, and what conformity and support conditions apply.
These research groups develop their own control, perception or motion logic. Documented SDK, ROS and simulation access matters more to them than pre-programmed demonstrations. RBTX describes development access for the G1 EDU, while Unitree provides its own development resources.
For experiments involving grasping, tactile feedback or human–robot interaction, the hand and sensor options need to suit the experiment. Unitree documents an optional Dex3-1 hand, while Generation Robots offers the Booster T1 in standard, gripper and hand variants.
If you’re planning dynamic movements near staff or test participants, you’ll need to set safety measures for the specific tasks and set-up. The DGUV recommends carrying out a risk assessment, including when choosing suitable emergency control devices.
Start with the experiments your lab actually plans to run. These will determine the software access, hardware options and safety measures you need. Then check how the specific configuration can be operated and supported throughout the project.
| Priority | What to check before buying | Who should pay particular attention |
|---|---|---|
| In-house development and simulation | Check SDK and API access, the ROS version supported, and the capabilities and limitations of the available simulation for the version on offer. | If we’re developing our own control or perception software, we need access to the interfaces required to do so. Around 36% of consumers put this development work first. |
| Sensors and manipulation | Confirm which sensors are included and specify the exact hand or gripper type, including its control system and any optional equipment. | For our grasping or interaction experiments, the hand supplied needs to suit the tasks. This is the top priority for around 29% of consumers. |
| Safe experiments | Carry out a risk assessment for the task and set-up. Use it to define the movement area, test zone and suitable emergency control devices. | If people are working near the robot, we need safety measures tailored to those specific experiments. Around 21% of consumers prioritise safe integration in the lab. |
| Operation throughout the project | Confirm in writing the availability of spare batteries and parts, repair arrangements, updates, support and conditions for remote access. | If an experiment is expected to run for a long time, we want to be able to respond to breakdowns or software changes. This kind of predictability comes first for around 14% of consumers. |
The development and hardware options described for the G1 EDU vary by version. Generation Robots lists separate standard, gripper and hand versions of the Booster T1, so the product family name alone doesn’t tell you what’s included.
Ask for the specific version, sensors, computing option, hand type, batteries and software access to be listed in your quote.
Additional hands, sensors or integration work can increase the project workload if they’re needed for your planned experiment.
Unitree notes that the Gazebo simulation described in its ROS repository does not handle high-level locomotion control. A simulation result also doesn’t prove that your specific lab set-up is safe.
Use simulation for suitable preliminary tests, then define safety measures and step-by-step approval criteria before running tests with the real robot.
Whether movements near people are permitted, and which emergency control device is suitable, depends on the task, machine and test set-up. The DGUV highlights this in relation to risk assessments.
Plan movement areas, people’s roles and emergency procedures before use, then use these to determine the safety measures required.
Your planned set-up may require designated test zones or additional safety equipment.
The Booster T1 listing described by Generation Robots is for a used demonstration model with limited availability. This alone doesn’t tell you how repairs, spare parts or updates will be handled during a research project.
Before ordering, agree in writing who to contact, how repairs will be arranged, how spare parts and batteries will be supplied, and what update and support conditions apply to the model on offer.
The G1 EDU family is worth considering for labs looking to test their own software and different hardware configurations.
The Booster T1 may be a suitable research option if interface access and the choice of a suitable gripper or hand are important.
The GR1 and GR2 are other listed humanoid robots that labs can consider when looking for a platform for embodied AI research.
No. Under Section 7 of Germany’s Product Safety Act, CE marking must be applied before a product is placed on the market where the relevant regulations require it. You’ll also need a task-specific risk assessment and suitable safety measures for use in your particular lab set-up, as the DGUV points out.
The risk assessment for your planned experiments should determine which emergency control device is suitable. The DGUV lists good visibility, quick access and ease of use as important requirements, and distinguishes between emergency stop and emergency switch-off functions.
What matters is the equipment included with the specific version on offer, not just the G1 EDU name. The RBTX Advanced-U2 page lists 43 degrees of freedom, 3D sensors and access for ROS, Python, C++ and Linux. The U1 page also describes development access, but notes that equipment options vary. Ask the supplier to confirm the sensors, computing module, hand and software access for the specific offer.