Jul 2026· 2026 IEEE/ASME International Conference on Advanced Intelligent Mechatronics (AIM)· pp. 1-6· 0 citations· 8 references
Abstract
This paper presents an end-to-end deep reinforcement learning (DRL) framework for integrated whole-body loco-manipulation control of a single-arm quadrupedal robot in contact-rich tasks. A single policy simultaneously controls all 18 joints of a Unitree Go2 robot equipped with a 6-DoF PiPER arm, trained in NVIDIA Isaac Lab using massively parallel simulation. The framework is evaluated on three contact-rich tasks: heavy-object dragging (up to 15 kg), heavy-object pushing, and elongated-object extraction from stacked configurations. The learned policy produces coordinated whole-body behaviors, where the legs provide propulsion and posture stabilization while the arm maintains task-oriented interaction with the object under strong contact forces. To investigate cross-simulator robustness, policies trained in Isaac Lab are directly evaluated in MuJoCo over 1,000 episodes under different domain randomization settings, showing that disturbance-aware training at the robot base achieves a 78.5% success rate in the dragging task and substantially outperforms friction and mass randomization alone.
Equipping quadruped robots with manipulators significantly expands their operational workspace. However, for small-scale systems constrained by limited joint torques, achieving robust whole-body control on unstructured terrains remains a substantial challenge. Existing learning-based methods often face an inherent trade-off between locomotion stability and manipulation dexterity: traversing terrains introduces continuous base perturbations that constantly disturb state observations, significantly disrupting precise manipulation learning, whereas training exclusively on flat ground fails to yield robust locomotion skills for unstructured environments. To address these challenges, we propose QLIMB, a novel end-to-end whole-body control framework tailored for small-scale quadruped manipulators. We introduce a latent belief mixing mechanism that adaptively fuses mode-specific state representations to decouple state estimation for agile locomotion and stable manipulation within a unified policy, enabling seamless transitions between mobility and interaction modes. Furthermore, the policy exhibits emergent leg-arm coordination, ensuring smooth postural adaptations and intrinsic self-balancing during manipulation. Extensive real-world experiments demonstrate that QLIMB enables small-scale quadruped manipulators to achieve robust locomotion and stable manipulation on challenging terrains.
Quancheng Qian, Peng Zhai, Zonghao Zhang et al.· IEEE Robotics and Automation...· 0 citations
This paper aims to present a hierarchical whole-body control (H-WBC) framework for humanoid shelf-picking in structured shelf environments. The objective is to improve motion coordination, posture regulation and safe task execution for humanoid manipulation in spatially constrained workspaces.
The proposed framework combines hierarchical whole-body kinematics, a unified task-constraint formulation and a strict-priority hierarchical quadratic programming scheme. End-effector tracking, waist posture regulation, arm-motion regularization, postural-stability constraints, joint-motion limits and shelf-related collision avoidance are integrated into a common velocity-level optimization framework. The method is evaluated in simulation and on the UBTECH Walker2 humanoid robot through single-arm and dual-arm shelf-picking tasks.
The proposed framework achieves accurate end-effector motion and coordinated whole-body behavior in constrained shelf environments. In simulation, it maintains low tracking errors in representative trajectory-following tasks, preserves postural stability and improves task success while reducing redundant arm motion in dual-arm shelf-picking across different shelf heights. In real-world experiments, it demonstrates practical execution of both single-arm pick-and-place and dual-arm box retrieval tasks, while maintaining stable and collision-aware whole-body motion near the shelf structure.
This study develops a reusable H-WBC framework tailored to humanoid shelf-picking. The proposed formulation unifies task objectives and physical constraints for shelf-picking within a strict-priority optimization hierarchy, and is validated in simulation and real-robot experiments on representative single-arm and dual-arm tasks in structured storage environments.
Xiao Li, Zhiyong Zhang, Lequn Fu et al.· Industrial robot· 0 citations
Effective learning design guidelines for realizing arm-based locomotion on life-sized robotic hardware and expanding the traversable workspace of robots are provided.
Ayumu Iwata, Kento Kawaharazuka, Keita Yoneda et al.· 0 citations
TAC-LOCO is proposed, a tactile-augmented unified reinforcement learning framework that encodes tactile array observations from compliant grippers into a compact latent representation and joins it with proprioception for unified control of the legs, arm, and gripper.
Muqun Hu, Yu-Hao Zhou, K. Malik et al.· arXiv.org· 0 citations
HAF (Humanoid Adaptation Framework), a two-part framework consisting of HAF-VLA and HAF-Steer that transfers off-the-shelf generalist VLA foundation models to humanoid whole-body loco-manipulation, surpasses vanilla single-stage VLA baselines and improves whole-body coordination and task performance.
Langzhe Gu, Chengkai Hou, Meng Li et al.· 0 citations
This paper introduces a new Kinematic Normalizing Flow (KNF) model, trained on a large-scale kinematic dataset, that generates diverse yet feasible partial reference motions that effectively addresses whole-body control challenges by decomposing the complex loco-manipulation problem into partial reference motion generation and low-level imitation control.
Zhengmao He, Moonkyu Jung, Hyeongjun Kim et al.· arXiv.org· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
