A team from the University of California Berkeley created a system that helps a multi-armed robot analyse items it needs to pick up and quickly determine which type of grip is the most likely to succeed.
The system uses a “reward function” assigned to each type of gripper which the robot then uses to quantify the probability of each type of grip succeeding on a particular item.
The developers said this enabled the system to rapidly decide which grip would most effectively pick up the object.
The researchers said in testing their two-armed robot had achieved a 95% reliability rate and suggested it could one day help speed up processing in e-commerce fulfilment centres.
Jeff Mahler, the research’s lead author and a postdoctoral researcher at the university said: “Any single gripper cannot handle all objects.
“For example, a suction cup cannot create a seal on porous objects such as clothing and parallel-jaw grippers may not be able to reach both sides of some tools and toys.”
The researchers argued that the system could be useful in warehouses in years to come, as companies increasingly turn to robotics to help fulfil growing order demand.