mlsanityData to 7 Oct 2026

RMA: Rapid Motor Adaptation for Legged Robots

arXiv 2107.04034 ↗First posted 8 Jul 2021cs.LG cs.AI cs.CVcited by 430

On the map: World models → Humanoid and legged locomotion world models

Abstract (arXiv)

Successful real-world deployment of legged robots would require them to adapt in real-time to unseen scenarios like changing terrains, changing payloads, wear and tear. This paper presents Rapid Motor Adaptation (RMA) algorithm to solve this problem of real-time online adaptation in quadruped robots. RMA consists of two components: a base policy and an adaptation module. The combination of these components enables the robot to adapt to novel situations in fractions of a second. RMA is trained completely in simulation without using any domain knowledge like reference trajectories or predefined foot trajectory generators and is deployed on the A1 robot without any fine-tuning. We train RMA on a varied terrain generator using bioenergetics-inspired rewards and deploy it on a variety of difficult terrains including rocky, slippery, deformable surfaces in environments with grass, long vegetation, concrete, pebbles, stairs, sand, etc. RMA shows state-of-the-art performance across diverse real-world as well as simulation experiments. Video results at https://ashish-kmr.github.io/rma-legged-robots/

Key facts. Cited by 430 papers in the map (data through 7 Oct 2026). Placed in Humanoid and legged locomotion world models, one of 11 directions in World models. Builds on 26 of its 70 references that are in the map. First posted on arXiv 8 Jul 2021.

Builds on

26 of 70 references.

References · 26

  1. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  2. 2017Proximal Policy Optimization Algorithmscited by 14,441
  3. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,861
  4. 2015Continuous control with deep reinforcement learningcited by 3,279
  5. 2018Addressing Function Approximation Error in Actor-Critic Methodscited by 2,302
  6. 2016Asynchronous Methods for Deep Reinforcement Learningcited by 2,218
  7. 2015High-Dimensional Continuous Control Using Generalized Advantage Estimationcited by 1,916
  8. 2017Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real Worldcited by 1,401
  9. 2017Sim-to-Real Transfer of Robotic Control with Dynamics Randomizationcited by 616
  10. 2020Learning Quadrupedal Locomotion over Challenging Terraincited by 494
  11. 2019Learning agile and dynamic motor skills for legged robotscited by 402
  12. 2018Sim-to-Real: Learning Agile Locomotion For Quadruped Robotscited by 277
14 more
  1. 2018Learning to Adapt in Dynamic, Real-World Environments Through Meta-Reinforcement Learningcited by 270
  2. 2020Learning Agile Robotic Locomotion Skills by Imitating Animalscited by 229
  3. 2018Learning to Walk via Deep Reinforcement Learningcited by 153
  4. 2017Preparing for the Unknown: Learning a Universal Policy with Online System Identificationcited by 129
  5. 2019Data Efficient Reinforcement Learning for Legged Robotscited by 60
  6. 2020Dynamics Randomization Revisited:A Case Study for Quadrupedal Locomotioncited by 54
  7. 2019Sim-to-Real Transfer for Biped Locomotioncited by 50
  8. 2019Policies Modulating Trajectory Generatorscited by 47
  9. 2019Learning Fast Adaptation with Meta Strategy Optimizationcited by 46
  10. 2020Rapidly Adaptable Legged Robots via Evolutionary Meta-Learningcited by 42
  11. 2018Policy Transfer with Strategy Optimizationcited by 41
  12. 2019Environment Probing Interaction Policiescited by 39
  13. 2019Multi-Agent Manipulation via Locomotion using Hierarchical Sim2Realcited by 39
  14. 2018Optimizing Simulations with Noise-Tolerant Structured Explorationcited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 12; 2022: 52; 2023: 75; 2024: 89; 2025: 89; 2026: 113.

430 papers.

Sort:

By area:

  1. 2023Eureka: Human-Level Reward Design via Coding Large Language Modelscited by 542
  2. 2024Evaluating Real-World Robot Manipulation Policies in Simulationcited by 484
  3. 2022DayDreamer: World Models for Physical Robot Learningcited by 372
  4. 2022ProcTHOR: Large-Scale Embodied AI Using Procedural Generationcited by 369
  5. 2023Orbit: A Unified Simulation Framework for Interactive Robot Learning Environmentscited by 239
  6. 2023RVT: Robotic View Transformer for 3D Object Manipulationcited by 224
  7. 2023Habitat 3.0: A Co-Habitat for Humans, Avatars and Robotscited by 220
  8. 2021A Survey of Zero-shot Generalisation in Deep Reinforcement Learningcited by 164
  9. 2024HumanPlus: Humanoid Shadowing and Imitation from Humanscited by 164
  10. 2022GNM: A General Navigation Model to Drive Any Robotcited by 161
  11. 2024Understanding World or Predicting Future? A Comprehensive Survey of World Modelscited by 161
  12. 2024Deep Reinforcement Learning for Robotics: A Survey of Real-World Successescited by 153
38 more
  1. 2023Structured World Models from Human Videoscited by 144
  2. 2024Open-TeleVision: Teleoperation with Immersive Active Visual Feedbackcited by 137
  3. 2023Human-Timescale Adaptation in an Open-Ended Task Spacecited by 136
  4. 2024Reconciling Reality through Simulation: A Real-to-Sim-to-Real Approach for Robust Manipulationcited by 135
  5. 2023Real-World Humanoid Locomotion with Reinforcement Learningcited by 132
  6. 2022Legged Locomotion in Challenging Terrains using Egocentric Visioncited by 131
  7. 2022Rapid Locomotion via Reinforcement Learningcited by 130
  8. 2023Extreme Parkour with Legged Robotscited by 120
  9. 2024Expressive Whole-Body Control for Humanoid Robotscited by 117
  10. 2023Robot Parkour Learningcited by 115
  11. 2022Walk These Ways: Tuning Robot Control for Generalization with Multiplicity of Behaviorcited by 113
  12. 2023Learning Agile Soccer Skills for a Bipedal Robot with Deep Reinforcement Learningcited by 111
  13. 2024Scaling Cross-Embodied Learning: One Policy for Manipulation, Navigation, Locomotion and Aviationcited by 108
  14. 2025ASAP: Aligning Simulation and Real-World Physics for Learning Agile Humanoid Whole-Body Skillscited by 104
  15. 2022In-Hand Object Rotation via Rapid Motor Adaptationcited by 102
  16. 2024HumanoidBench: Simulated Humanoid Benchmark for Whole-Body Locomotion and Manipulationcited by 101
  17. 2023Toward General-Purpose Robots via Foundation Models: A Survey and Meta-Analysiscited by 99
  18. 2022Deep Whole-Body Control: Learning a Unified Policy for Manipulation and Locomotioncited by 98
  19. 2022Deep Hierarchical Planning from Pixelscited by 96
  20. 2022Visual Dexterity: In-Hand Reorientation of Novel and Complex Object Shapescited by 92
  21. 2023General In-Hand Object Rotation with Vision and Touchcited by 87
  22. 2024From Imitation to Refinement -- Residual RL for Precise Assemblycited by 80
  23. 2022MetaMorph: Learning Universal Controllers with Transformerscited by 78
  24. 2023LEAP Hand: Low-Cost, Efficient, and Anthropomorphic Hand for Robot Learningcited by 78
  25. 2024DrEureka: Language Model Guided Sim-To-Real Transfercited by 78
  26. 2022A Walk in the Park: Learning to Walk in 20 Minutes With Model-Free Reinforcement Learningcited by 76
  27. 2024ExBody2: Advanced Expressive Humanoid Whole-Body Controlcited by 75
  28. 2026PointWorld: Scaling 3D World Models for In-The-Wild Robotic Manipulationcited by 74
  29. 2022Adversarial Motion Priors Make Good Substitutes for Complex Reward Functionscited by 68
  30. 2021Minimizing Energy Consumption Leads to the Emergence of Gaits in Legged Robotscited by 67
  31. 2024Reinforcement Learning for Versatile, Dynamic, and Robust Bipedal Locomotion Controlcited by 63
  32. 2023Gen2Sim: Scaling up Robot Learning in Simulation with Generative Modelscited by 62
  33. 2022Concurrent Training of a Control Policy and a State Estimator for Dynamic and Robust Legged Locomotioncited by 61
  34. 2024Visual Whole-Body Control for Legged Loco-Manipulationcited by 58
  35. 2025Visual Imitation Enables Contextual Humanoid Controlcited by 57
  36. 2021Robot Learning from Randomized Simulations: A Reviewcited by 56
  37. 2024Agile But Safe: Learning Collision-Free High-Speed Legged Locomotioncited by 53
  38. 2024AdaptiGraph: Material-Adaptive Graph-Based Neural Dynamics for Robotic Manipulationcited by 52

50 most cited of 430 All 430 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2023Learning Terrain-Adaptive Locomotion with Agile Behaviors by Imitating Animals
  2. 2023A Real-World Quadrupedal Locomotion Benchmark for Offline Reinforcement Learning
  3. 2026LocoWM: High-Precision Locomotion through World-Model-Guided Residual Adaptation
  4. 2021Learning Coordinated Terrain-Adaptive Locomotion by Imitating a Centroidal Dynamics Planner
  5. 2025GenTe: Generative Real-world Terrains for General Legged Robot Locomotion Control
  6. 2024Spatio-Temporal Motion Retargeting for Quadruped Robots
  7. 2023Manipulator as a Tail: Promoting Dynamic Stability for Legged Locomotion
  8. 2020Dynamics and Domain Randomized Gait Modulation with Bezier Curves for Sim-to-Real Legged Locomotion