At CES 2026 in Las Vegas, Google DeepMind and Boston Dynamics unveiled a new phase in robotics by integrating DeepMind’s Gemini AI model into advanced robotic platforms, starting with Boston Dynamics’ humanoid Atlas, tells Wired.com. This collaboration aims to give robots the contextual awareness and decision-making abilities necessary to operate in dynamic industrial environments rather than follow rigid, preprogrammed instructions. The effort marks a shift from robots that can perform scripted acrobatics to machines capable of interpreting complex surroundings and manipulating objects.
Atlas, long known for its mobility and balance, exemplifies Boston Dynamics’ decades-long work in dynamic robots, from hydraulically actuated humanoids to fully electric machines with an enhanced range of motion. With Gemini, the goal is to equip Atlas with multimodal intelligence that combines perception, reasoning, and adaptable control, allowing it to navigate unfamiliar settings and perform tasks that were previously out of reach for industrial robots.
Testing set to begin at Hyundai’s automotive factories signals the shift toward real-world application. Gemini-powered Atlas units will attempt manufacturing tasks such as sorting parts and handling materials on actual production floors rather than controlled research settings, an important step toward broad deployment. Hyundai, which owns Boston Dynamics, anticipates that this blend of robotics hardware and AI software will transform operational capabilities in factories and beyond.
The partnership underscores a broader trend in robotics research: moving AI out of the purely digital realm and into physical systems where perception, manipulation, and decision-making converge. Gemini’s role is not merely to execute predefined moves but to interpret sensor data, identify objects, and respond intelligently to changing conditions, that is, capabilities that, if successful at scale, could expand robots’ usefulness far beyond traditional automation.
Safety remains a priority. DeepMind and Boston Dynamics emphasize robust safeguards to prevent unsafe behavior as autonomous robots begin working alongside humans in real environments. This next chapter in industrial robotics highlights the intersection of advanced AI models with complex mechanical systems, pointing toward machines capable of learning and adapting in ways that bring them closer to human-like versatility.