ZhilianhaiGalbot: G1 Mobile Dual-Arm Robots and Embodied Intelligence
Company and product identity
Galbot traces its establishment to 2023. Its official company overview describes an embodied-intelligence business spanning robot hardware, models and training data. Galbot G1 and S1 are product families within this company. AstraBrain and AstraData identify its model and data work; they are not additional robot manufacturers.
G1 physical architecture
The current G1 product specification describes a four-wheel omnidirectional mobile base carrying two seven-axis arms and an adjustable upper body. This is a mobile manipulation platform, rather than a bipedal walking robot. The table lists a wrist-centre payload of 5 kg per arm, 10 kg combined. Tooling, object dimensions and the proposed motion still need configuration-specific assessment.
Configuration and operating limits
The current product table lists IP54 protection and an operating temperature range of 0–40 °C. Its wrist-centre workspace should be read separately from broader promotional reach descriptions. Confirm the exact model revision, chosen end effectors and environmental limits in the delivered documentation. An eight-hour listed operating time is not evidence of uninterrupted performance under every load or duty cycle.
Versioned development tools
Galbot publishes a G1 user guide and versioned developer documentation. The C++ SDK reference covers joint and trajectory control, end-effector operation, mobile-base velocity, sensor access and coordinate transformations. Match the SDK, robot software and selected machine type before integration. Public APIs provide development building blocks; they do not establish that a finished application or connector is supplied.
Application evidence and trials
The manufacturer presents retail, warehouse and industrial handling applications. Evaluate the proposed task with actual objects, shelves and access constraints, including difficult surfaces, obstructed views and interrupted picks. Record successful cycles together with operator assistance, failed attempts and recovery time. Broad embodied-intelligence claims should be tested against a defined task and acceptance dataset before making production commitments.
Site and workflow integration
Map travel routes, docking positions and human work areas before a trial. Clarify how orders reach the robot, how completion is reported and who resolves a failed operation. Test the combination of navigation, reaching and grasping in the proposed layout. A successful stationary pick does not establish the performance of an entire mobile workflow.
Handover and support questions
Request version records, operator training, configuration backups and a written support scope. Agree responsibility for changing tools, adding objects and maintaining sensors or charging equipment. For rollout decisions, compare accepted throughput, intervention rate and service arrangements across the full shift. Keep prototype demonstrations, commissioned installations and any future availability statements distinct in the purchase record.