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

arXiv: https://arxiv.org/abs/2205.01179 · HTML: https://mlsanity.com/paper/2205.01179/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 44 references that are in the map.
- First posted on arXiv 2 May 2022.

## Builds on (most cited first, 22 of 22)

- [Auto-Encoding Variational Bayes](https://mlsanity.com/paper/1312.6114/) (2013)
- [Learning Latent Dynamics for Planning from Pixels](https://mlsanity.com/paper/1811.04551/) (2018)
- [Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)](https://mlsanity.com/paper/1511.07289/) (2015)
- [Learning agile and dynamic motor skills for legged robots](https://mlsanity.com/paper/1901.08652/) (2019)
- [Embed to Control: A Locally Linear Latent Dynamics Model for Control from Raw Images](https://mlsanity.com/paper/1506.07365/) (2015)
- [Character Controllers Using Motion VAEs](https://mlsanity.com/paper/2103.14274/) (2021)
- [Deep Variational Bayes Filters: Unsupervised Learning of State Space Models from Raw Data](https://mlsanity.com/paper/1605.06432/) (2016)
- [RLOC: Terrain-Aware Legged Locomotion using Reinforcement Learning and Optimal Control](https://mlsanity.com/paper/2012.03094/) (2020)
- [Fast and Efficient Locomotion via Learned Gait Transitions](https://mlsanity.com/paper/2104.04644/) (2021)
- [Universal Planning Networks](https://mlsanity.com/paper/1804.00645/) (2018)
- [Guided Constrained Policy Optimization for Dynamic Quadrupedal Robot Locomotion](https://mlsanity.com/paper/2002.09676/) (2020)
- [Multi-Layered Safety for Legged Robots via Control Barrier Functions and Model Predictive Control](https://mlsanity.com/paper/2011.00032/) (2020)
- [Vision-aided Dynamic Quadrupedal Locomotion on Discrete Terrain using Motion Libraries](https://mlsanity.com/paper/2110.00891/) (2021)
- [An Efficient Optimal Planning and Control Framework For Quadrupedal Locomotion](https://mlsanity.com/paper/1609.09861/) (2016)
- [Planning in Learned Latent Action Spaces for Generalizable Legged Locomotion](https://mlsanity.com/paper/2008.11867/) (2020)
- [Receding-Horizon Perceptive Trajectory Optimization for Dynamic Legged Locomotion with Learned Initialization](https://mlsanity.com/paper/2104.09078/) (2021)
- [Agile Maneuvers in Legged Robots: a Predictive Control Approach](https://mlsanity.com/paper/2203.07554/) (2022)
- [First Steps: Latent-Space Control with Semantic Constraints for Quadruped Locomotion](https://mlsanity.com/paper/2007.01520/) (2020)
- [A Feasibility-Driven Approach to Control-Limited DDP](https://mlsanity.com/paper/2010.00411/) (2020)
- [Fast Trajectory Optimization for Legged Robots using Vertex-based ZMP Constraints](https://mlsanity.com/paper/1705.10313/) (2017)
- [Model Predictive Control with Environment Adaptation for Legged Locomotion](https://mlsanity.com/paper/2105.05998/) (2021)
- [Next Steps: Learning a Disentangled Gait Representation for Versatile Quadruped Locomotion](https://mlsanity.com/paper/2112.04809/) (2021)

## Built on by (most cited first, 2 of 2)

- [Gaitor: Learning a Unified Representation Across Gaits for Real-World Quadruped Locomotion](https://mlsanity.com/paper/2405.19452/) (2024)
- [Agile perceptive multi-skill locomotion for quadrupedal robots in the wild](https://mlsanity.com/paper/2607.13579/) (2026)

## Abstract (arXiv)

Quadruped locomotion is rapidly maturing to a degree where robots are able to realise highly dynamic manoeuvres. However, current planners are unable to vary key gait parameters of the in-swing feet midair. In this work 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 synthesising 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 synthesised 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 whilst being robust and reactive to external perturbations.