Learning Reactive Dexterous Grasping via Hierarchical Task-Space RL Planning and Joint-Space QP Control

Figure 1 : Overview of the proposed multi-agent RL framework with hardware validation. Our manipulator platform, equipped with a five-finger anthropomorphic hand, successfully grasps and lifts a target object. The background overlay illustrates the policy decomposition. The arm agent \pi_{\text{arm}} (red) commands the palm twist (red and orange arrows), while the hand agent \pi_{\text{hand}} (blue) independently commands the local fingertip linear velocities (light blue arrows). The bottom row highlights successful grasps on arbitrary object shapes and deformables.

Figure 2 : Overview of the hybrid hierarchical control framework. Our architecture consists of a high-level RL planner operating at 100 Hz and a low-level QP controller running at 500 Hz. The planner is decomposed into two specialized agents: (i) an arm agent that generates a global transport strategy via a desired palm twist {}^{W}\boldsymbol{\mathcal{V}}_{P,\text{des}} in the world frame, and (ii) a hand agent that generates local grasping strategy via desired fingertip linear velocities {}^{P}\mathbf{v}_{F,\text{des}/P} relative to the palm. This decomposition enables a unified global-to-lo

Figure 3 : Illustration of the fixed-based manipulation platforms and coordinate frames. Left: Two-fingered gripper. Right: Five-fingered anthropomorphic hand. Zoomed views show the local palm frame P used for fingertip velocity mapping and the computation of hand-centric local observations. The target object is modeled as a minimum volume bounding box (green), enabling the framework to generalize across various arbitrary object shapes.

Figure 5 : Evolution of the arm and hand policy action outputs over time corresponding to the toy airplane execution sequence shown in Fig. 4 (a). Phase transitions are intrinsically driven by the decreasing palm-to-object distance (bottom). During reaching, the palm rapidly approaches the object while the fingers remain relatively passive. During grasping, the palm linear velocity drops for precise positioning while the fingertip velocities peak to actively enclose the object. In the lifting stage, the palm translates to the desired pose while the fingers maintain continuous velocity commands

Figure 6 : Per-object grasp success rates for the 5F hand, highlighting the five lowest-performing (left) and highest-performing (right) objects. The first line of the x-axis labels denotes the corresponding YCB object ID. The results demonstrate that geometrically uniform objects accommodate diverse grasping strategies, yielding higher success rates, whereas objects with extreme aspect ratios or asymmetric geometries demand highly precise and specialized grasps that are harder to execute.

Figure 12 : Spatial distribution of the palm velocity tracking error across the task-space Cartesian planes. The colormap illustrates the mean tracking error magnitude, while the dashed line outlines the mean joint velocity limit mapped into the operational space. The region of high tracking accuracy (white) spatially conforms to the geometric shape of the joint velocity limit envelope. This spatial alignment confirms that the tracking performance is strictly governed by the physical capabilities of the manipulator, demonstrating that the error growth observed in Fig. 11 is a direct consequenc