We have been accustomed to the fact that Boston Dynamics engineers show off several times a year what their humanoid robots have learned. While five years ago, they had to be held by stabilizing ropes while walking, today they can easily do gymnastics and parkour.
Of course, Boston dynamics is not the only robotic studio. More and more bizarre skills are being taught to humanoids and bipedal machines by a bunch of other robotics companies, as well.
One such a robot is UC Berkeley’s Cassie Cal, a meter-long limb without a large body above it. This two-legged ostrich-looking robot was recently taught to juggle while maintaining balance.
The whole routine is entirely self-contained and autonomous. The video shows that the juggling does not involve hands; instead, the robot is bouncing a ball on what would be its head, while maintaining balance to avoid falling over.
Cassie’s juggling is assisted by an external motion capture system that tracks the position of the court and the ball. The control algorithm then allows Cassie to juggle the ball by leaning forward and backward, left and right, and moving up and down. Most importantly, Cassie Cal does this while maintaining the balance, which was the whole point of this research.
The main goal behind the research is to explore dynamic multitasking, a skill that robots will need in order to be successful in human environments, the team explains.
Engineers from the University of California at Berkeley in 2017 introduced the two-legged robot Cassie and then created on its base the humanoid robot Digit, which differs from Cassie in the presence of the body and arms. For two years, developers have constantly been teaching robots various private non-trivial skills, suggesting that in the future, these developments can be used to create a universal robot that can be put into practice.