Solving a Rubik’s cube is a challenge many of us first attempt in childhood. Only a few go on to master the deceptively trickly puzzle, while the rest of us remain utterly stumped by the colourful conundrum.
Robots, on the other hand, find solving Rubik’s cubes an absolute cinch – at least in theory. Physically manipulating the cube to perform the twists and turns required to complete the puzzle is another matter altogether.
Researchers at the University of Washington have capitalised on the difficulty of this fine motor challenge to design a protocol to first train and then test robots on their ability to manipulate physical objects. Publishing in IEEE Robotics and Automation Letters, the team describe why solving a Rubik’s cube is the ultimate task to probe the mismatch between what humans and robots find challenging, and begin to bridge that gap.
“For us humans, it is challenging to decide what moves to make to complete Rubik’s cube, yet it is quite easy for us to physically manipulate and twist the cube as desired,” says study author Dr Boling Yang from the Paul G Allen School of Computer Science and Engineering at the University of Washington, US.
“For robots, it is the other way around. Robots only need a few seconds to execute a search-based algorithm that can determine a sequence of moves that will solve the cube. However, executing the individual manipulations in the real world requires sub-centimetre positioning accuracy.”
The trouble lies in the long sequence of manipulations needed to solve the cube. If an error is made in one movement, this can affect subsequent movements. As errors build on errors, the overall performance begins to degrade.
For humans, this is easy to pick up on and account for by adjusting our movements. But for robots, this fine-tuning based on continuous assessment is difficult. Those that can pull it off, particularly those that can do it quickly, can therefore be assumed to have high manipulation performance.
The protocol that the researchers present assesses both the speed and accuracy of a robot’s cube-solving efforts, with robots receiving the highest scores if they perform well in both areas.
“Robots can score highly on our protocol by successfully completing a large number of manipulations and/or minimising the time required to perform manipulations,” Yang explains. “Our protocol thus establishes a baseline to which other researchers can compare their systems.”
Having tested their protocol on two well-known but distinct robotic platforms – PR2 and Herb – the researchers are confident that their protocol is general enough to be applied to a variety of different robots, and could be used to evaluate different computational methods that underpin robotic manipulation.
And there’s an added bonus – it’s a whole lot of fun.
“With respect to a more general audience, Rubik’s cube is understandable and even enjoyable to watch by almost everyone, thus it could serve as a gateway for students to be introduced to robotics,” says Yang. “We are now interested in inviting researchers specialised in different areas, such as sensing, planning, and manipulator design, to try our benchmark and use it to advance robot manipulation in terms of accuracy and speed.”