Jumpin Cube: How Does It Really Work?

This little Cube Robot was published in 2015. They call this a “soft robot”, though that is really a misuse of the term, since the drive mechanisms are “hard”.   As far as I can tell, the soft materials only serve to make the locomotion by bouncing work better.

(There is also a lot of cargo cult science in the paper, in the form of trivial equations and a pointless experiment.  This looks like an undergraduate exercise, which is probably was.)

On the other hand, this is a simple, sort of bio-inspired drive mechanism, using a flat spring to jump. This concept isn’t necessarily all that new, though. I think it would work much better at smaller scales (molecular machines already do this) or in low gravity, such as on Mars or an Asteroid or Comet.

What caught my attention this week is the question of how it is steered. One video is titles “A Soft Robot Cube Capable of Controllable Jumping”. The video shows the cube bouncing around and effectively ascending and then descending a small mound. The captions says “Goal: move bottom to top” and “Goal: move from top to bottom” (Falling down the hill was much quicker than climbing.)

There is a definite implication that the cube is somehow steering toward a goal, directing the hops to try to move in the intended direction. The press releases claim that the robot is “autonomous”, though the paper says that future work include “design a directional jumping algorithm together with a closed-loop control strategy for the robot”. This seems to imply that the demonstration might be steered by a preprogrammed algorithm, or even remote control.

What we see in the video shows the robot apparently selecting which slot to emit the spring from, thus selecting the direction of the hop. So how does it know which slot to choose? Going “up hill”, it picks a slot that is under the cube and pointing downhill. The jump is only approximately in the right direction, and the cube bounces and rolls. So, how does it know its position for he second hop?

For that matter, how does it know where the top of the hill is? Or which way is “up” or “down”?

I admit that I am baffled. The information I have is long on “look what I built”, and short on how it actually works.


  1. Shuguang Li, Robert Katzschmann, and Daniela Rus. A soft cube capable of controllable continuous jumping. In 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2015, 1712-1717.

 

Robot Wednesday

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s