GUANGZHOU, Feb 2: Forget the clunky, rigid robots of today. Researchers in China have unveiled a new type of soft, lightweight humanoid robot designed to be a safe and adaptable helper in future homes, Xinhua reported.
Inspired by “growable” human bones, the robot named GrowHR is 1.36 metres tall but weighs only 4.5 kg, which is lighter than a large house cat. A six-year-old child can easily lift GrowHR.
This extreme lightness, combined with a body made of flexible, air-filled structures make it safer than current heavy, unyielding robots, drastically reducing the risk of injury or damage if it bumps into people or furniture.
Using air pressure, GrowHR can shrink when needing to pass through tight spaces, like squeezing under a low table or through a narrow hallway – abilities crucial for navigating cluttered homes.
This robotic prototype can also deform to walk through a low tunnel that is only 36 per cent of its original height, and can squeeze through a narrow gap that is just 61 per cent of its original width by contracting its belly, according to the team from Southern University of Science and Technology in south China’s Shenzhen. The team have published details of their innovation in the journal Science Advances.
GrowHR’s soft body ensures safe human interaction, allowing hugging, falling, and lifting without injury. Additionally, the robot is compact and easy to transport. It can be packed into a small box for storage.
The robot’s lightweight design enables other unique functions. Notably, it can float and swim, suggesting potential for tasks such as retrieving items from a pool.
Its low weight also means it can be carried by a drone to a location 5.5 km away, hinting at future delivery or remote-assistance applications.
This work pioneers a growable, multifunctional robotic design approach for dynamic, complex environments, the researchers said.
The team acknowledges future room for improvement that includes increased degrees of freedom, enhanced dynamic performance, and greater autonomy enabled by advanced control strategies and learning algorithms, such as large language models.
— BERNAMA-XINHUA