mlsanityData to 14 Oct 2026

Gait in Eight: Efficient On-Robot Learning for Omnidirectional Quadruped Locomotion

arXiv 2503.08375 ↗First posted 11 Mar 2025cs.RO cs.LGcited by 8

On the map: Deep reinforcement learning methods → RL for robot locomotion and manipulation

Abstract (arXiv)

On-robot Reinforcement Learning is a promising approach to train embodiment-aware policies for legged robots. However, the computational constraints of real-time learning on robots pose a significant challenge. We present a framework for efficiently learning quadruped locomotion in just 8 minutes of raw real-time training utilizing the sample efficiency and minimal computational overhead of the new off-policy algorithm CrossQ. We investigate two control architectures: Predicting joint target positions for agile, high-speed locomotion and Central Pattern Generators for stable, natural gaits. While prior work focused on learning simple forward gaits, our framework extends on-robot learning to omnidirectional locomotion. We demonstrate the robustness of our approach in different indoor and outdoor environments.

Key facts. Cited by 8 papers in the map (data through 14 Oct 2026). Placed in RL for robot locomotion and manipulation, one of 15 directions in Deep reinforcement learning methods. Builds on 26 of its 41 references that are in the map. First posted on arXiv 11 Mar 2025.

Builds on

26 of 41 references.

Closest ancestors · 1

Same direction, newest first

  1. 2024One Policy to Run Them All: an End-to-end Learning Approach to Multi-Embodiment Locomotion10 Sep

Other references · 25

  1. 2017Proximal Policy Optimization Algorithmscited by 15,215
  2. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 7,005
  3. 2018Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actorcited by 4,885
  4. 2017Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real Worldcited by 1,498
  5. 2021Learning to Walk in Minutes Using Massively Parallel Deep Reinforcement Learningcited by 512
  6. 2021Randomized Ensembled Double Q-Learning: Learning Fast Without a Modelcited by 288
  7. 2022The Primacy Bias in Deep Reinforcement Learningcited by 247
  8. 2022Rapid Locomotion via Reinforcement Learningcited by 168
  9. 2023Extreme Parkour with Legged Robotscited by 164
  10. 2022Legged Locomotion in Challenging Terrains using Egocentric Visioncited by 151
  11. 2023Robot Parkour Learningcited by 149
  12. 2021Dropout Q-Functions for Doubly Efficient Reinforcement Learningcited by 114
13 more
  1. 2019CrossQ: Batch Normalization in Deep Reinforcement Learning for Greater Sample Efficiency and Simplicitycited by 95
  2. 2022Concurrent Training of a Control Policy and a State Estimator for Dynamic and Robust Legged Locomotioncited by 88
  3. 2022A Walk in the Park: Learning to Walk in 20 Minutes With Model-Free Reinforcement Learningcited by 83
  4. 2022Learning Free Gait Transition for Quadruped Robots via Phase-Guided Controllercited by 29
  5. 2023Grow Your Limits: Continuous Improvement with Real-World RL for Robotic Locomotioncited by 24
  6. 2021Robust High-speed Running for Quadruped Robots via Deep Reinforcement Learningcited by 22
  7. 2023Learning Agile Locomotion on Risky Terrainscited by 21
  8. 2025Scaling Off-Policy Reinforcement Learning with Batch and Weight Normalizationcited by 19
  9. 2024Learning to Walk from Three Minutes of Real-World Data with Semi-structured Dynamics Modelscited by 16
  10. 2024Stage-Wise Reward Shaping for Acrobatic Robots: A Constrained Multi-Objective Reinforcement Learning Approachcited by 13
  11. 2023RL-X: A Deep Reinforcement Learning Library (not only) for RoboCupcited by 8
  12. 2021Aggressive Q-Learning with Ensembles: Achieving Both High Sample Efficiency and High Asymptotic Performancecited by 7
  13. 2021A CPG-Based Agile and Versatile Locomotion Framework Using Proximal Symmetry Losscited by 6

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 5; 2026: 3.

8 papers.

  1. 2025XQC: Well-conditioned Optimization Accelerates Deep Reinforcement Learningcited by 17
  2. 2025Towards Embodiment Scaling Laws in Robot Locomotioncited by 16
  3. 2025Multi-Embodiment Locomotion at Scale with extreme Embodiment Randomizationcited by 4
  4. 2026Shape Your Body: Value Gradients for Multi-Embodiment Robot Designcited by 2
  5. 2025Bridge the Gap: Enhancing Quadruped Locomotion with Vertical Ground Perturbationscited by 1
  6. 2026Rapid Embodiment Adaptation for Quadrupedal Locomotioncited by 1
  7. 2025From Tabula Rasa to Emergent Abilities: Discovering Robot Skills via Real-World Unsupervised Quality-Diversitycited by 0
  8. 2026OmniMimic: Dynamics-completed Motion Augmentation for Multi-style Omnidirectional Quadruped Locomotioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025GRoQ-LoCO: Generalist and Robot-agnostic Quadruped Locomotion Control using Offline Datasets
  2. 2023A Real-World Quadrupedal Locomotion Benchmark for Offline Reinforcement Learning
  3. 2026Learning Gait-Aware Quadruped Locomotion with Temporal Logic Specifications
  4. 2024Offline Adaptation of Quadruped Locomotion using Diffusion Models
  5. 2025Energy-Efficient Omnidirectional Locomotion for Wheeled Quadrupeds via Predictive Energy-Aware Nominal Gait Selection
  6. 2025Acrobotics: A Generalist Approach to Quadrupedal Robots' Parkour
  7. 2026Agile perceptive multi-skill locomotion for quadrupedal robots in the wild
  8. 2025Training Directional Locomotion for Quadrupedal Low-Cost Robotic Systems via Deep Reinforcement Learning