close
close

The robot hand can detach from the arm, crawl up to objects and pick them up

The robot hand can detach from the arm, crawl up to objects and pick them up

Research videos are not known for their story arcs. Sometimes, though, you see one that just keeps getting better the more you watch. “Beyond Manual Dexterity” is one such video. He impresses you from the first few seconds with a very dexterous robotic pincer, then things start to go off the rails.

The video was first presented alongside a research paper of the same name at the IEEE International Conference on Robotics and Automation in Rotterdam this week. The work, led by EPFL in Switzerland, explores ways in which robotic hands can overcome existing limitations to grasp an even wider range of objects.

“The deep learning model has significantly advanced dexterous manipulation techniques for multi-fingered hand grasping,” the team notes. “However, contact information-guided typing in cluttered environments remains largely underexplored.”

For decades, roboticists have drawn inspiration from the natural world. Today’s robots are built largely on these foundations. Look no further than the quadrupeds and humanoids currently flooding the market. Organic is a rich and seemingly inexhaustible source of inspiration, but it can also push designers into a bind.

But why must a robotic hand be constrained by the limitations of its natural counterpart? The prevailing answer is perhaps a lack of imagination. The research presented here, however, shows what these systems look like when we intentionally break these connections.

The results can be a little scary. Design a human-inspired hand that can reach back to pick up a wide range of objects and detach itself to crawl towards another, planting it firmly inside the uncanny valley. If it didn’t look so much like a human hand, it would probably be less disturbing.

Image credits: EPFL (4x speed)

Operating in this world, however, requires pushing through this visceral reaction. Beyond that, you’ll find a path forward to progress in robotics. Whether something like this can be scaled successfully is another question. For now, the research is undoubtedly generating a lot of interest in the field.

“Even though you see it in horror movies,” IEEE researcher Xiao Gao said cheekily, “I think we are the first to introduce this idea into robotics.”

Goodbye biologically inspired robots; hello horror robots.