ISO 13482 is entering final approval at the moment domestic humanoid makers are shipping products to real households. The standard addresses hazard identification and risk assessment but stops short of enforceable rules for the safety challenges that emerge from human-robot relationships, according to Jae-Seong Lee of South Korea's Electronics and Telecommunications Research Institute.
The core technical problem is that home robot safety is bidirectional. The robot changes human behavior, and humans change what the robot perceives and does next. Safety is not a fixed machine property — it emerges from the relationship. Current standards acknowledge non-contact hazards like unpredictability and incorrect autonomous decisions but provide no binding test methods or compliance criteria.
Domestic humanoids must operate in uncontrolled environments with elderly residents, children, visitors, pets, and clutter. Tightening the operating envelope to factory-like conditions would eliminate the value proposition. Meanwhile, training datasets for these robots now include real-world variability captured by contract workers recording household chores globally, meaning robots will encounter the full spectrum of human states and behaviors.
The standards gap is governance. Lee argues that older adults — the primary intended users of domestic care robots — are not systematically represented in ISO working groups. The result: safety assumptions get embedded in products before markets, regulators, and users can challenge them. Once deployment patterns harden, revising the baseline becomes much harder.
The revision pushes unresolved questions into advisory language rather than enforceable requirements. The question it defers: what counts as safe relational behavior when the human is part of the system, not background noise?



