Dynamics Aware Quadrupedal Locomotion via Intrinsic Dynamics Head

• URL: http://arxiv.org/abs/2605.01227v1
• Date: Sat, 02 May 2026 03:50:54 GMT
• Title: Dynamics Aware Quadrupedal Locomotion via Intrinsic Dynamics Head
• Abstract summary: We propose a training framework for quadrupedal locomotion that enables the Control Policy to understand and reason about physical dynamics.<n>In simulation, we concurrently train an Intrinsic Dynamics (ID) Head that learns state-to-torque dynamics alongside the Control Policy.<n>We show that this mechanism drives convergence to better optima across a wide range of standard quadrupedal locomotion rewards, yielding more efficient and smoother policies.
• Score: 0.5672132510411465
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Quadrupedal locomotion plays a critical role in enabling agile, versatile movement across complex terrains. Understanding and estimating the underlying physical dynamics are essential for achieving efficient and stable quadrupedal locomotion. We propose a novel training framework for quadrupedal locomotion that enables the Control Policy to understand and reason about physical dynamics. In simulation, we concurrently train an Intrinsic Dynamics (ID) Head that learns state-to-torque dynamics alongside the Control Policy, and we define a dynamics reward enabled by the ID Head that encourages the Policy toward more predictable dynamical behavior. We also provide a mechanism to tune the learned dynamics in the resulting Policy by controlling the training coefficients of the ID Head. Our simulation experiments show that this mechanism drives convergence to better optima across a wide range of standard quadrupedal locomotion rewards, yielding more efficient and smoother policies. Our real-robot experiments demonstrate sim-to-real transfer of these improvements, with significant gains in torque efficiency (16.8%), action rate (18.6%), and mechanical power (12.8%), while improving safe torque occupancy by 6.4%.

Related papers

• Extending the Speed Limit of Quadrupedal Locomotion via Refined Actuator Modeling and Adaptive Command Scheduling [0.9298381080776307]
  We propose a high-speed locomotion framework that reduces sim-to-real discrepancies and stabilizes learning over a wide command distribution.<n>A refined actuator model explicitly captures high-speed voltage coupling and magnetic saturation, enabling a more accurate representation of the torque-speed envelope.<n>Experiments on the 36.5 kg quadruped BlackPanther2 demonstrate speeds of up to 13.2 m/s on a treadmill and 11.65 m/s outdoors, establishing a new state-of-the-art and, to the best of our knowledge, a world record for quadrupedal robot locomotion.
  arXiv  Detail & Related papers  (2026-09-09T08:45:38Z)
• KYON: Semi-Modular Wheel-Legged Quadruped With Agile Bimanual Capability [3.041420062518373]
  KYON is a wheel-legged quadruped robot equipped with a bimanual upper body for loco-manipulation tasks.<n>The platform features a semi-modular design with a reconfigurable lower legs, enabling both wheeled and legged locomotion depending on the environment.
  arXiv  Detail & Related papers  (2026-06-29T12:54:09Z)
• Learning Dynamic Pick-and-Place for a Legged Manipulator [8.027246382283003]
  This work presents a hierarchical reinforcement learning framework for dynamic pick-and-place tasks using a quadruped equipped with a 6-DOF robotic arm.<n>In simulation, the system achieves an 86.05% success rate with payloads up to 2.3 kg.<n>Within a wide vertical workspace ranging from ground level to 1.1m-high tabletops, the system demonstrates an average success rate of 73.3% for payloads up to 1.3 kg.
  arXiv  Detail & Related papers  (2026-05-15T08:04:07Z)
• Articulated-Body Dynamics Network: Dynamics-Grounded Prior for Robot Learning [14.046146512352315]
  We present the Articulated-Body Dynamics Network (ABD-Net), a novel graph neural network architecture grounded in the computational structure of forward dynamics.<n>Our approach demonstrates increased sample efficiency and generalization to dynamics shifts compared to transformer-based and GNN baselines.<n>We further validate the learned policy on real Unitree G1 and Go2 robots, state-of-the-art humanoid and quadruped platforms.
  arXiv  Detail & Related papers  (2026-03-19T16:06:49Z)
• X-Loco: Towards Generalist Humanoid Locomotion Control via Synergetic Policy Distillation [50.0220259870317]
  We introduce X-Loco, a framework for training a vision-based generalist humanoid locomotion policy.<n>X-Loco trains multiple oracle specialist policies and adopts a synergetic policy distillation with a case-adaptive specialist selection mechanism.<n>This design enables the student to acquire a broad spectrum of locomotion skills.
  arXiv  Detail & Related papers  (2026-03-04T05:07:05Z)
• HUSKY: Humanoid Skateboarding System via Physics-Aware Whole-Body Control [67.58927286226925]
  We propose a learning-based framework that integrates humanoid-skateboard system modeling and physics-aware whole-body control.<n>Experiments on the Unitree G1 humanoid platform demonstrate that our framework enables stable and agile maneuvering on skateboards in real-world scenarios.
  arXiv  Detail & Related papers  (2026-02-03T07:18:01Z)
• Reinforcement Learning for Versatile, Dynamic, and Robust Bipedal Locomotion Control [106.32794844077534]
  This paper presents a study on using deep reinforcement learning to create dynamic locomotion controllers for bipedal robots. We develop a general control solution that can be used for a range of dynamic bipedal skills, from periodic walking and running to aperiodic jumping and standing. This work pushes the limits of agility for bipedal robots through extensive real-world experiments.
  arXiv  Detail & Related papers  (2024-01-30T10:48:43Z)
• Exploring Model Transferability through the Lens of Potential Energy [78.60851825944212]
  Transfer learning has become crucial in computer vision tasks due to the vast availability of pre-trained deep learning models. Existing methods for measuring the transferability of pre-trained models rely on statistical correlations between encoded static features and task labels. We present an insightful physics-inspired approach named PED to address these challenges.
  arXiv  Detail & Related papers  (2023-08-29T07:15:57Z)
• Active Learning of Discrete-Time Dynamics for Uncertainty-Aware Model Predictive Control [46.81433026280051]
  We present a self-supervised learning approach that actively models the dynamics of nonlinear robotic systems. Our approach showcases high resilience and generalization capabilities by consistently adapting to unseen flight conditions.
  arXiv  Detail & Related papers  (2022-10-23T00:45:05Z)
• Learning Low-Frequency Motion Control for Robust and Dynamic Robot Locomotion [10.838285018473725]
  We demonstrate robust and dynamic locomotion with a learned motion controller executing at as low as 8 Hz on a real ANYmal C quadruped. The robot is able to robustly and repeatably achieve a high heading velocity of 1.5 m/s, traverse uneven terrain, and resist unexpected external perturbations.
  arXiv  Detail & Related papers  (2022-09-29T15:55:33Z)
• Learning to Exploit Elastic Actuators for Quadruped Locomotion [7.9585932082270014]
  Spring-based actuators in legged locomotion provide energy-efficiency and improved performance, but increase the difficulty of controller design. We propose to learn model-free controllers directly on the real robot. We evaluate the proposed approach on the DLR elastic quadruped bert.
  arXiv  Detail & Related papers  (2022-09-15T09:43:17Z)
• Reinforcement Learning for Robust Parameterized Locomotion Control of Bipedal Robots [121.42930679076574]
  We present a model-free reinforcement learning framework for training robust locomotion policies in simulation. domain randomization is used to encourage the policies to learn behaviors that are robust across variations in system dynamics. We demonstrate this on versatile walking behaviors such as tracking a target walking velocity, walking height, and turning yaw.
  arXiv  Detail & Related papers  (2021-03-26T07:14:01Z)
• Neural Dynamic Policies for End-to-End Sensorimotor Learning [51.24542903398335]
  The current dominant paradigm in sensorimotor control, whether imitation or reinforcement learning, is to train policies directly in raw action spaces. We propose Neural Dynamic Policies (NDPs) that make predictions in trajectory distribution space. NDPs outperform the prior state-of-the-art in terms of either efficiency or performance across several robotic control tasks.
  arXiv  Detail & Related papers  (2020-12-04T18:59:32Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.