Humanoid robots are poised to make a significant leap into commercial viability, according to recent announcements from Chinese and US companies planning large-scale production. The past five years have witnessed a marked improvement in humanoid capabilities, fueled by cost reductions in component parts and advancements in battery technology and artificial intelligence algorithms, enabling enhanced perception and autonomy.

UBTECH, a Chinese firm, announced in November that it had achieved the world's first mass delivery of humanoid robots. More than 1,000 of its Walker S2 model humanoids were shipped to factories with the expectation of deployment in 2025, according to Yu Zheng, a roboticist and vice-dean of the UBTECH Research Institute in Shenzhen. Zheng described the Walker S2 as capable of autonomous and stable walking, as well as object manipulation, while noting that deployment remained in its initial phases.

The potential for humanoids to deliver tangible benefits to companies in terms of time or cost savings is still under evaluation. Current limitations include battery life measured in hours, and the continued reliance on human operators who, in many instances, remotely control the robots to execute tasks while gathering data. This mirrors early stages of industrial automation, reminiscent of the 1960s when robotic arms first entered automotive assembly lines, initially performing only repetitive welding tasks under close supervision.

The automotive industry is already experimenting with humanoid robots. Trials are underway to assess their suitability for various manufacturing processes. This echoes the historical trajectory of industrial robots, which gradually expanded their roles from simple, repetitive tasks to more complex operations as technology advanced.

The advancements in AI algorithms are crucial, providing the "brainpower" for these robots to perceive their environment and make autonomous decisions. This represents a significant upgrade from earlier generations of robots that relied heavily on pre-programmed instructions. However, the extent to which these robots can truly operate independently and efficiently in real-world scenarios remains to be seen.