VAE-Loco: Versatile Quadruped Locomotion by Learning a Disentangled Gait Representation

Quadruped locomotion is rapidly maturing to a degree where robots are able to realize highly dynamic maneuvers. However, current planners are unable to vary key gait parameters of the in-swing feet midair. In this article, we address this limitation and show that it is pivotal in increasing controller robustness by learning a latent space capturing the key stance phases constituting a particular gait. This is achieved via a generative model trained on a single trot style, which encourages disentanglement such that application of a drive signal to a single dimension of the latent state induces holistic plans synthesizing a continuous variety of trot styles. We demonstrate that specific properties of the drive signal map directly to gait parameters, such as cadence, footstep height, and full-stance duration. Due to the nature of our approach, these synthesized gaits are continuously variable online during robot operation. The use of a generative model facilitates the detection and mitigation of disturbances to provide a versatile and robust planning framework. We evaluate our approach on two versions of the real ANYmal quadruped robots and demonstrate that our method achieves a continuous blend of dynamic trot styles while being robust and reactive to external perturbations.

Automated summary

In this article, the limitation of current quadruped locomotion planners is addressed, and it is shown that varying key gait parameters of the in-swing feet midair is crucial for increasing controller robustness. A generative model trained on a single trot style is used to learn a latent space capturing the key stance phases constituting a particular gait, enabling holistic plans synthesizing a continuous variety of trot styles. It is demonstrated that drive signal properties directly map to gait parameters, and these synthesized gaits are continuously variable online during robot operation, facilitated by the generative model. The approach is evaluated on two versions of the real ANYmal quadruped robots, achieving a continuous blend of dynamic trot styles while being robust and reactive to external perturbations.