On July 30, 2026, Google announced the Gemini Robotics 2 platform, emphasizing its new intelligent whole-body control capability. The reveal included a demonstration video that showed robots performing simple household chores such as picking up trash and lifting watering cans.
The platform integrates sensor data with motion algorithms to synchronize the movement of all limbs and the torso as a single controllable unit. By continuously adjusting joint angles based on real‑time feedback, the robot can shift its weight and reposition its arms while maintaining balance. This coordinated control lets the machine reach for objects on the ground or adjust its grip without losing stability.
The system uses embedded cameras and inertial sensors to build a live model of the robot’s posture. Control software then calculates the necessary torque for each joint to achieve a desired task while keeping the center of mass within a safe envelope. This approach treats the whole body as one actuator, allowing smooth transitions between locomotion and manipulation.
In the footage, robots navigate a test zone, locate scattered pieces of litter, and pick them up using a gripper. Another sequence shows a robot approaching a table, grasping a watering can, and lifting it upright without spilling. These actions illustrate the platform’s ability to combine movement with precise object handling.
Demonstrations suggest that Gemini Robotics 2 could be useful in settings where robots must move freely while interacting with objects, such as homes, offices, or warehouses. By merging navigation and manipulation into one control loop,