TAIPEI (Taiwan News) — Ho Hsin-yu (何心宇), a researcher at the Institute for Information Industry’s Market Intelligence & Consulting Institute, said that as humanoid robot technology gradually matures, its multi-tasking, collaborative, and flexible capabilities are set to broaden applications from smart manufacturing to everyday life.
Ho predicted that within the next 20 years, humanoid robots could account for up to 60% of the global robotics market, working alongside other automation and industrial robots.
Since Tesla unveiled its Optimus humanoid robot in 2021, its development has progressed rapidly. Last year, the company showcased the Optimus Gen 2, which stands 173 cm tall and weighs 56 kg. It is capable of walking at 8 kph, carrying heavy objects, gently gripping an egg, and navigating rugged outdoor terrain.
The robot also demonstrated stable ball-catching, highlighting its significantly improved dexterity. Tesla utilized its EV technology, including batteries, actuators, sensor systems, and autonomous driving technology, to enhance the robot's stability.
Nvidia has also expanded into robotics. In March 2024, it launched Project Generalist Robot 00 Technology (GR00T), an initiative aimed at advancing robotics and integrating AI technology. According to Nvidia, GR00T-powered robots can understand human language and imitate human movements through observation.
As part of the project, Nvidia also introduced Jetson Thor, a new computer designed for humanoid robots, and upgraded its Isaac robotics platform. These improvements include AI tools that help robots learn, simulate tasks, and better understand human actions.
Figure AI, backed by investors including Nvidia, unveiled its latest humanoid robot, the Figure 03, earlier this month. The new model features an advanced sensor suite and a redesigned hand control system, cutting manufacturing costs while enhancing usability in everyday environments.
Figure 03 has fewer parts and a streamlined assembly compared with its previous version. The robot is powered by Figure AI’s Helix AI model, which enables autonomous decision-making. Its upgraded sensors and hand design allow the robot to handle precise tasks in tight spaces.
Despite global R&D efforts, Ho noted two main challenges remain in the sector: software complexity, which requires high training costs and large data sets, and hardware limitations, as dexterous hand control demands precise movement, strong load capacity, and remains expensive.
Some industry operators also noted that humanoid robots still lag behind traditional industrial automation in stability and precision. ASEA Brown Boveri said commercial adoption is far off, while Yaskawa emphasized mimicking human work over appearance. Boston Dynamics and Collaborative Robotics added that they are not ideal for all tasks.
Ho highlighted that smart manufacturing, particularly in the automotive and ICT sectors, will be the first environment for widespread humanoid robot use. As the technology advances, he predicts such robots will expand into homes, commercial spaces, schools, and the food service industry.
Source: Institute for Information Industry’s Market Intelligence & Consulting Institute (MIC)
Author: MIC Researcher Ho Hsin-yu
