To build their robot, the team started with stick insect designs. They then modified them in ways that mimicked the dung beetle. Then, instead of using machine learning to teach the robot what to do, as is done with many other modern robots, the researchers used a modular, neural-based, loco-manipulation control approach based on just two behaviors—pitch and roll for the upper two pairs of legs and simple biomechanics for the lower pair. The result was a robot they call ALPHA.
At last, we can build a better, stronger Sisyphus, one who can roll a boulder up a hill 1000x more efficiently.
Sisyphus, now jobless, can be denied unemployment and socioeconomically rot from having his slightly above minimum wage income taken away by a robot.