Anyone who has placed their hands on their laps and are to pick up a coffee mug on the table would obviously extract their hands from under the table before stretching their arm out toward the mug.
But unfortunately for robots, performing the same task involves crunching a lot of mathematical calculations to get the work done.
Now, researchers have stumbled upon a shortcut which could make robots take that small step towards being human – at least when it comes to holding a coffee cup.
Current motion-planning algorithms take too long to figure out the most efficient path from one position to another, so these robots decide to just find a path that will not cause the robot’s limbs to smack into the table.
This gets the job done, but can lead to peculiar sets of motion that waste time and energy, in addition to potentially frightening any humans around it.
Matthew Walter, a computer scientist at MIT, and his team have developed a new algorithm that takes a different approach.
Their robot will instead assume that a sphere of open space surrounds each point it evaluates, hence all points within this space will be ignored, reports New Scientist.
As the robot maps out all potential collisions, it will shrink or expand these spheres accordingly, resulting in an overlapping volume of clear space, which allows it to move through naturally and freely.