Singapore engineers build robotic fingers with delicate grip | ZDNet

Engineers from National University of Singapore (NUS) have built a robotics system they say can grip various objects, ranging from soft and delicate to bulky and heavy. Designed to be configurable, the robotic hand is touted to address the needs of sectors such as vertical farming, food assembly, and fast-moving consumer goods packaging, and with a 23% improvement in efficiency.

These industries increasingly were automating more of their operations, but currently required manual handling for some processes, according to NUS. The human hand’s natural dexterity remained necessary for these tasks. 

Rave Yeow, associate professor from NUS Advanced Robotics Centre and Department of Biomedical Engineering, said: “An object’s shape, texture, weight, and size affect how we choose to grip them. This is one of the main reasons why many industries still heavily rely on human labour to package and handle delicate items.”

To address this, the robotic grippers can be built with three or four 3D-printed fingers that are soft and flexible and sit on a reconfigurable base. The fingers are air-driven and armed with a locking mechanism that enables the level of stiffness to be adjusted. 

The robotic system is guided by computer vision and deep learning technologies, so it can identify the type of objects and determine its orientation. It then automatically decides how it can best pick and place the objects to reduce the need for heavy human intervention. 

“Our hybrid robotic gripper technology revolutionises traditional pick-and-place tasks by offering advanced capabilities that allow robots to safely interact with delicate items of various shapes, sizes and stiffness, just like the human hand,” said Yeow, who led the project. 

According to NUS, the gripper system is configurable “on demand” and can be fitted with three grip options that cater to different uses.

The GourmetGrip, for instance, is fit for granular tasks such as handling delicate bite-sized food items or products that are prone to damage, including tofu, and packing them into takeout boxes. Described as a “soft-handed mode”, this option can support different grip posses and operate within various space restrictions. 

Mounted on an industrial robotic arm, GourmetGrip could pick and place food items at a speed that was comparable to a human’s, and with consistency round the clock, NUS said. The university added that this system mode delivered a 23% improvement in gripping efficiency compared to other commercially available grippers, holding items faster and with higher precision.

GourmetGrip currently is able to pick up more than 50 types of food items, including pudding, sliced cakes, and fruits.

Another gripper option, Universal Soft Gripper (UnisoGrip) is suitable for handling packaged goods along assembly lines, in particular, when these products are at the final stages of being packed into boxes for shipping. This option, which is expected to be the most flexible in terms of task application, can be  substantially expanded in its grip range, NUS said. The system is larger compared to GourmetGrip and handle items that measure up to 30cm wide and weigh up to 3kg. 

UnisoGrip also has rotatable soft gripper fingers that are suitable for delicate products and a vacuum suction cup that enables the system to navigate tighter spaces or awkwardly placed projects, such as the corner of a tote bin. To date, it can pick up more than 30 types of consumer goods including coffee powder packs, refillable detergent packs, and bottled drinks.

Customers also can build a gripping system based on both GourmetGrip and UnisoGrip, and fully customised to their business requirements and space constraints, according to NUS. In fact, a system was tailored for Singapore-based manufacturer, People Bee Hoon Factory, which produces rice vermicelli. It commissioned the NUS team to build a system that was optimised for packing rice vermicelli packets into carton boxes.

People Bee Hoon Factor’s director Desmond Goh said: “Most of our existing staff are mature workers, so we sought to tap new technology that can ease the workload of our existing staff, while simultaneously boosting their productivity. We selected this technology because it is able to meet our purpose and provides flexibility for different deployments that we require.”

The NUS engineers, through their startup RoPlus (RO+), now are looking to commercialise the product. They are pitching to offer potential commercial partners various options of implementing the robotic system, either bundled with a robotic arm or sold individually to customers that already own their own commercial robot arms, which then can be mounted to the gripper system. 

A complete product offering comprising the gripper, robotic arm, computer vision capabilities, and conveyor system also will be available. 

The NUS engineers currently are working to expand the system’s ability to recognise more items, using machine learning and an online library to train the system.

RO+ is funded under the NUS Graduate Research Innovation Programme, which seeds startups founded by the university’s academic staff and researchers. The team includes researchers Low Jin Huat, Khin Phone May, and Chen Chao-Yu, as well as undergraduate student Han Qian Qian.

Research that drove the gripper system also was supported by Singapore’s National Robotics Programme and Agency for Science, Technology and Research.

RELATED COVERAGE