Quadrupedal robots with manipulators offer strong mobility and adaptability for grasping in unstructured, dynamic environments through coordinated whole-body control. However, existing research has predominantly focused on static-object grasping, neglecting the challenges posed by dynamic targets and thus limiting applicability in dynamic scenarios such as logistics sorting and human–robot collaboration. To address this, we introduce DQ-Bench, a new benchmark that systematically evaluates dynamic grasping across varying object motions, velocities, heights, object types, and terrain complexities, along with comprehensive evaluation metrics. Building upon this benchmark, we propose DQ-Net, a compact teacher–student framework designed to infer grasp configurations from limited perceptual cues. During training, the teacher network leverages privileged information to holistically model both the static geometric properties and dynamic motion characteristics of the target, and integrates a grasp fusion module to deliver robust guidance for motion planning. Concurrently, we design a lightweight student network that performs dual-viewpoint temporal modeling using only the target mask, depth map, and proprioceptive state, enabling closed-loop action outputs without reliance on privileged data. Extensive experiments on DQ-Bench demonstrate that DQ-Net achieves robust dynamic objects grasping across multiple task settings, substantially outperforming baseline methods in both success rate and responsiveness. We will release our codebase and benchmark publicly.

In the RoboCup 3D simulation competition, how to find a flexible and stable gait pattern is one of the keys to win the match. To achieve such walking gait, a machine learning method of optimizing the vertical Center of Mass(CoM) trajectory is presented. The vertical CoM trajectory is planned by multiple polynomial function. Inverted Pendulum Model(IPM) and a numerical method are utilized to control the Zero Moment Point(ZMP). Then the key parameters are extracted from the gait pattern, a distributed multi-robot training environment based on RoboCup 3D simulated platform is constructed, and the parallel multi-swarm particle swarm algorithm is applied to optimize the parameters. The results of experiment and competition demonstrate that the effectiveness of the proposed method.