Major tech companies are investing heavily in so-called physical AI: humanoid robots capable of operating autonomously in the real world. But how close are we to seeing such robots in our homes? Three researchers from the Amsterdam University of Applied Sciences (HvA) offer their perspective.
According to Merel Noorman, Associate Professor of Responsible AI at HvA, the idea of humanoid robots has circulated for decades. While robotics has made substantial progress, particularly in industrial settings, large-scale future visions often outpace practical reality.
Pascal Wiggers, Professor of Responsible IT, does not expect humanoid robots to become commonplace in households or workplaces in the near future. Domestic environments are fundamentally unpredictable. Objects move, people and pets create dynamic situations, and materials vary in weight and texture. Robots must continuously adapt while avoiding physical damage.
Historically, robots perform best in carefully controlled environments. Factories are structured to accommodate robotic systems, often following serious safety incidents. Such controlled conditions do not exist in everyday household settings.
Progress is more visible in social robotics. Tibert Verhagen, Professor of Emerging Technology for Business, notes that retailers are increasingly experimenting with robots designed to assist customers, provide product advice and answer questions. Powered by advanced language models, these systems communicate more fluently than before. Research at HvA indicates that customers value robots particularly for functional, routine tasks.
Nevertheless, fundamental challenges remain. Noorman highlights the so-called “frame problem” in AI: the physical world contains too many variables to be fully modelled. Unexpected situations are inevitable. Wiggers adds that AI systems require extensive training data, yet mistakes during the learning phase of physical robots can result in tangible harm.
Miscommunication further complicates matters. Humans constantly clarify intentions and repair misunderstandings in conversation. Robots may fail to grasp context or nuance. Even text-based AI systems can cause harm through misinterpretation; physical robots introduce an additional safety dimension.
The researchers conclude that while social robots will likely continue to expand in specific sectors, a reliable humanoid robot assistant in the home remains a distant prospect.