However, there’s still one thing our artificial friends aren’t so good at: tasks that involve a gentle touch. Chances are, if you give something delicate to a robot, it’ll crush the object in its hands in seconds.
Now, in a surprising step forward, scientists at the ISI Laboratory at the University of Tokyo have successfully used machine learning to teach its robots how to peel soft fruit—like a banana—without squishing it.
According to Laughing Squid, while handling soft materials is still a difficult task for most robots, the models in question were able to develop the motor skills needed to peel a banana after hours of training with a human operator.
As per FutureTimeline, the actual task is rather arduous and requires a total of nine steps, such as: grasping the banana, picking it off the table, locating the tip, peeling it, and adjusting the grip so that all of the skin can be removed.
In total, the robot took three minutes to execute the process, so you’re probably still a lot quicker.
Take a look at the video below and witness the incredible dexterity of the robot at work.