We are searching data for your request:
Upon completion, a link will appear to access the found materials.
Taking its example from us mere mortals, a new method of using dual arm robot systems has been created. Rather than working as independent pieces, these dual armed robots work as part of a coherent whole. Just like humans.
The incredible result was a robot performing breakfast-making duties more coherently than if each of its parts were being individually controlled. By accurately mimicking humans and our two-handed methods of motion, these robots’ skillset could expand exponentially.
RELATED: NOVEL STUDY EXPLORES HOW ROBOTS GRAB THINGS FROM HUMANS
Starting small, yet impressively, this could mean that robots assist humans with our everyday, menial tasks in the future. Anything from opening a can to whisking up the laundry basket to the other side of the house.
Could dual-armed robots serve us breakfast in bed in the future?
This phenomenon, known as the ‘gestalt’ effect, meaning arms and hands coordinating and working together in order to achieve what each individual limb cannot create on its own, has already been critical in human society and its advancements. These critical advancements include tool use, manual labor, meal preparation, and communicative gestures.
How do humans do this without thinking?
The main theory is that humans use specialized sections of our brain that hold centralized symbolic representations of dual motions. Authors of the research, Daniel Ratika and colleagues, took this as their inspiration when creating a ‘bimanual action vocabulary’’, which they took from extensive human bimanual hand and arm motions analysis.
How did researchers train their robots?
The researchers took a robot that was programmed with a neural network (just like our brains) as well as a bimanual action vocabulary which was linked to 24 different volunteers. Each of these volunteers wore motion capture gloves.
Over the span of 10 minutes, each of the participants acted out 15 different breakfast-making motions, such as pouring juice or cracking eggs. During this time, the robot captured the humans’ poses and re-enacted the correct motion by going through its bimanual vocabulary.
The most surprising and important part of it all? The robot was able to independently override the humans’ motion commands and create a sequence of movements to perform the task more efficiently. In doing so, the robot completed the tasks more successfully than if it were to follow action-by-action human movements.
Next up: Breakfast served faster, by two-handed robots.