In this paper, we propose a robust controller that achieves
natural and stably fast locomotion on a real blind quadruped robot. With
only proprioceptive information, the quadruped robot can move at a
maximum speed of 10 times its body length per second, and has the
ability to pass through various complex terrains. The controller is
trained in the simulation environment by model-free reinforcement
learning and no specific human knowledge such as a foot trajectory
generator are used in the training architecture. Our research finds that
there is a problem of data symmetry loss during training, which leads to
unbalanced performance of the learned controller on the left-right
symmetric quadruped robot structure, and proposes a mirror-world neural
network to solve the performance problem. The learned controller
composed of the mirror-world network can make the robot achieve
excellent anti-disturbance ability. The learned controller can
coordinate the robot’s gait frequency and locomotion speed, and has good
generalization ability to reach locomotion speed it has never learned
and traverse terrains it has never seen before.