Scientists from MIT have created a robot hand system that can reorient over 2,000 different types of objects. The technique combines artificial intelligence (AI) with training to program the hand, according to a recent paper published on the preprint server ArXiv. It’s part of a growing effort to develop robot hands that resemble those of people. “These hands are highly dexterous and capable of performing in-hand manipulation,” Carmel Majidi, a professor of Mechanical Engineering and director of the Soft Machines Lab at Carnegie Mellon University’s College of Engineering, who was not involved in the paper, told Lifewire in an email interview. “That is, in addition to grasping and releasing objects, they can use their fingers to manipulate an object like a screwdriver or scissors.”
Better Handiwork
Making robot hands with human abilities presents a formidable challenge. Scientists from MIT say their invention can manipulate anything from a cup to a tuna can to a Cheez-It box, and it could help the hand quickly pick and place objects in specific ways and locations, The new techniques could assist in logistics and manufacturing, helping with typical demands such as packing objects into slots or manipulating a broader range of tools. The team used a simulated, anthropomorphic hand with 24 degrees of freedom and showed evidence that the system could be transferred to a real robotic system in the future. “In commercial applications, a parallel-jaw gripper is most commonly used, partially due to its simplicity in control, but it’s physically unable to handle many tools we see in daily life,” Tao Chen, the lead researcher on the project, said in a news release. “Even using a plier is difficult because it can’t dexterously move one handle back and forth. Our system will allow a multi-fingered hand to dexterously manipulate such tools, which opens up a new area for robotics applications.” Shenli Yuan, a research engineer at SRI International’s Robotics Laboratory, said in an email to Lifewire that it’s hard to create robotic hands that mimic the capabilities of humans because they have such great dexterity. He noted that human hands are anatomically complex, with many muscles, bones, tendons, and ligaments involved in each motion. “They are also packed with mechanoreceptors that provide us with rich haptic feedback,” Yuan added. “Most importantly, dexterity does not come from hands alone, and it is also very much related to our abilities to comprehend the environment and plan for the tasks we are performing.” While the advancement in robotic hands has been going on for over a century, “we still don’t have actuators comparable to human muscles in terms of similar force density and efficiency, sensors with similar fidelity and coverage compared to the tactile sensor on our hands, or the same level of intelligence to perform general tasks,” Yuan said.
Future Functions
The development of robotic hands is advancing rapidly, Yuan said. For example, many non-anthropomorphic robot hands are being designed to provide capabilities that exceed human hands. There has been a lot of work on tactile sensors that can give robot hands very high-fidelity tactile feedback. Continued research could lead to more specialized robot hands handling a more comprehensive range of industrial tasks, Yuan said. He predicted that the tasks robots could perform would get increasingly complex. “However, we might not see anthropomorphic hands in the factories anytime soon because, most likely, there will be simpler and more efficient hand designs depending on the tasks,” Yuan added. “In the long term, if robots get deployed into our homes or offices, we might see certain robot end-effectors that better resemble human hands because these environments are very much designed around human interaction [and] needs.” Many robotic picking companies like Berkshire Grey employ vacuum-based grippers, which are easier to use and currently more capable than finger-based grippers. Christopher Geyer, an engineer at the company, told Lifewire via email that the system could transform the supply chain. “Whereas the globalization of goods was, in large part, due to the automation of shipping containers, automation of unit handling is going to lower the cost of goods much more locally,” he added.
