mlsanityData to 7 Oct 2026

SoloParkour: Constrained Reinforcement Learning for Visual Locomotion from Privileged Experience

arXiv 2409.13678 ↗First posted 20 Sep 2024cs.RO cs.CV cs.LGcited by 10

Not on the map. Why?

Abstract (arXiv)

Parkour poses a significant challenge for legged robots, requiring navigation through complex environments with agility and precision based on limited sensory inputs. In this work, we introduce a novel method for training end-to-end visual policies, from depth pixels to robot control commands, to achieve agile and safe quadruped locomotion. We formulate robot parkour as a constrained reinforcement learning (RL) problem designed to maximize the emergence of agile skills within the robot's physical limits while ensuring safety. We first train a policy without vision using privileged information about the robot's surroundings. We then generate experience from this privileged policy to warm-start a sample efficient off-policy RL algorithm from depth images. This allows the robot to adapt behaviors from this privileged experience to visual locomotion while circumventing the high computational costs of RL directly from pixels. We demonstrate the effectiveness of our method on a real Solo-12 robot, showcasing its capability to perform a variety of parkour skills such as walking, climbing, leaping, and crawling.

Key facts. Cited by 10 papers in the map (data through 7 Oct 2026). Builds on 62 of its 77 references that are in the map. First posted on arXiv 20 Sep 2024.

Builds on

62 of 77 references.

References · 62

  1. 2017Proximal Policy Optimization Algorithmscited by 14,441
  2. 2016Layer Normalizationcited by 5,579
  3. 2015Continuous control with deep reinforcement learningcited by 3,279
  4. 2014Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modelingcited by 2,341
  5. 2021Isaac Gym: High Performance GPU-Based Physics Simulation For Robot Learningcited by 882
  6. 2020Learning Quadrupedal Locomotion over Challenging Terraincited by 494
  7. 2021RMA: Rapid Motor Adaptation for Legged Robotscited by 430
  8. 2019Learning agile and dynamic motor skills for legged robotscited by 402
  9. 2021Learning to Walk in Minutes Using Massively Parallel Deep Reinforcement Learningcited by 395
  10. 2022DayDreamer: World Models for Physical Robot Learningcited by 372
  11. 2023Efficient Online Reinforcement Learning with Offline Datacited by 332
  12. 2022Learning robust perceptive locomotion for quadrupedal robots in the wildcited by 299
38 more
  1. 2021Randomized Ensembled Double Q-Learning: Learning Fast Without a Modelcited by 278
  2. 2020Learning Agile Robotic Locomotion Skills by Imitating Animalscited by 229
  3. 2017Asymmetric Actor Critic for Image-Based Robot Learningcited by 202
  4. 2022Legged Locomotion in Challenging Terrains using Egocentric Visioncited by 131
  5. 2022Rapid Locomotion via Reinforcement Learningcited by 130
  6. 2023Extreme Parkour with Legged Robotscited by 120
  7. 2023Robot Parkour Learningcited by 115
  8. 2023ANYmal Parkour: Learning Agile Navigation for Quadrupedal Robotscited by 105
  9. 2024DrEureka: Language Model Guided Sim-To-Real Transfercited by 78
  10. 2022A Walk in the Park: Learning to Walk in 20 Minutes With Model-Free Reinforcement Learningcited by 76
  11. 2021Blind Bipedal Stair Traversal via Sim-to-Real Reinforcement Learningcited by 73
  12. 2021Minimizing Energy Consumption Leads to the Emergence of Gaits in Legged Robotscited by 67
  13. 2021Legged Robots that Keep on Learning: Fine-Tuning Locomotion Policies in the Real Worldcited by 64
  14. 2024Reinforcement Learning for Versatile, Dynamic, and Robust Bipedal Locomotion Controlcited by 63
  15. 2020Multi-expert learning of adaptive legged locomotioncited by 62
  16. 2024Agile But Safe: Learning Collision-Free High-Speed Legged Locomotioncited by 53
  17. 2024DiffuseLoco: Real-Time Legged Locomotion Control with Diffusion from Offline Datasetscited by 47
  18. 2019DeepGait: Planning and Control of Quadrupedal Gaits using Deep Reinforcement Learningcited by 45
  19. 2023Neural Volumetric Memory for Visual Locomotion Controlcited by 40
  20. 2021Learning to Jump from Pixelscited by 39
  21. 2022Advanced Skills by Learning Locomotion and Local Navigation End-to-Endcited by 39
  22. 2023Barkour: Benchmarking Animal-level Agility with Quadruped Robotscited by 33
  23. 2023Not Only Rewards But Also Constraints: Applications on Legged Robot Locomotioncited by 33
  24. 2022Learning Agile Skills via Adversarial Imitation of Rough Partial Demonstrationscited by 30
  25. 2023Learning Vision-Based Bipedal Locomotion for Challenging Terraincited by 30
  26. 2022Safe Reinforcement Learning for Legged Locomotioncited by 29
  27. 2021VILENS: Visual, Inertial, Lidar, and Leg Odometry for All-Terrain Legged Robotscited by 25
  28. 2023Learning and Adapting Agile Locomotion Skills by Transferring Experiencecited by 25
  29. 2022Learning Visual Locomotion with Cross-Modal Supervisioncited by 23
  30. 2017Learning Ground Traversability from Simulationscited by 20
  31. 2021Robust High-speed Running for Quadruped Robots via Deep Reinforcement Learningcited by 20
  32. 2022OPT-Mimic: Imitation of Optimized Trajectories for Dynamic Quadruped Behaviorscited by 19
  33. 2019An Open Torque-Controlled Modular Robot Architecture for Legged Locomotion Researchcited by 18
  34. 2020Robust Quadruped Jumping via Deep Reinforcement Learningcited by 18
  35. 2022Creating a Dynamic Quadrupedal Robotic Goalkeeper with Reinforcement Learningcited by 17
  36. 2023Generalized Animal Imitator: Agile Locomotion with Versatile Motion Priorcited by 17
  37. 2021Model-free Reinforcement Learning for Robust Locomotion using Demonstrations from Trajectory Optimizationcited by 16
  38. 2021Learning Inertial Odometry for Dynamic Legged Robot State Estimationcited by 16

50 most cited of 62 All 62 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 2; 2025: 4; 2026: 4.

10 papers.

  1. 2025Towards Embodiment Scaling Laws in Robot Locomotioncited by 12
  2. 2024Student-Informed Teacher Trainingcited by 6
  3. 2025First Order Model-Based RL through Decoupled Backpropagationcited by 3
  4. 2025QuietPaw: Learning Quadrupedal Locomotion with Versatile Noise Preference Alignmentcited by 1
  5. 2025Bipedalism for Quadrupedal Robots: Versatile Loco-Manipulation through Risk-Adaptive Reinforcement Learningcited by 1
  6. 2026Constraint-Aware Diffusion Priors for High-Fidelity and Versatile Quadruped Locomotioncited by 1
  7. 2024Reinforcement Learning from Wild Animal Videoscited by 0
  8. 2026Coupled Local and Global World Models for Efficient First Order RLcited by 0
  9. 2026Co-jump: Cooperative Jumping with Quadrupedal Robots via Multi-Agent Reinforcement Learningcited by 0
  10. 2026LEAP: Learning Emergent Active Perception for Quadruped Navigationcited by 0