Technology
Why home robots need wheels, not legs—and a smarter brain to match
As humanoids flood tech shows, one reviewer argues the real future of domestic robotics lies in wheeled bots with coordinated AI, not expensive machines that do everything poorly.
The Verge
Humanoid robot housekeepers have long felt just out of reach—a promise of the future that never quite arrives. But 2025 is shaping up differently. Companies like 1X have already deployed legged robots into homes, while dozens of human-shaped competitors are hitting tech show runways worldwide this month. Yet after years of living with various domestic robots, the evidence suggests we're building toward the wrong vision.
The case against full humanoids#
The appeal of a humanoid butler is obvious: one machine that handles everything. The reality is messier. Homes present chaotic, unpredictable environments that differ wildly from factories, where humanoids have found genuine footing. Most American homes include stairs, which demand either flying capability or powerful leg strength—and neither works well for a living space. When challenged to simply open a laundry detergent bottle, humanoid robots at CES uniformly failed.
More fundamentally, tech history teaches a consistent lesson: gadgets rarely master multiple functions well. A machine designed to vacuum, fold clothes, cook, and clear clutter becomes an expensive generalist—good at none of those tasks.
The smarter alternative: a robot brain#
Rather than one hulking machine attempting everything, the path forward involves an AI-powered coordinator that orchestrates existing devices based on household conditions and needs. Such a system could dispatch a robot vacuum when the dog spills kibble, time laundry for cheap electricity, or send a lawnmower out when grass needs trimming. It would reason and adapt, not merely follow pre-set routines.
LG's CLOiD embodies this hybrid approach: a wheeled platform with seven-degree-of-freedom arms and individually actuated fingers, paired with smart home controls. The company is now training hundreds of CLOiDs at a Seoul data factory to generate roughly 100,000 hours of training footage by year's end—feeding a proprietary foundation model built on Nvidia's open-world tools.
Tuya is pursuing a different angle with Doova, an AI home companion designed for seniors. Rather than tackle laundry or dishes, it tracks occupants, enables emergency video calls with family, provides conversation, and manages connected appliances—all without humanoid complexity.
