Search papers, labs, and topics across Lattice.
The paper introduces HierKick, a dual-frequency hierarchical reinforcement learning framework for vision-guided soccer robot control that decomposes the task into high-level tactical decisions and low-level motor control. This hierarchical approach addresses the challenges of multi-time-scale decision-making in complex environments, where end-to-end RL often struggles. The framework achieves high success rates (95.2% in IsaacGym, 89.8% in Mujoco, and 80% in the real world) by using a 5 Hz high-level policy with YOLOv8 object detection and a 50 Hz pre-trained low-level controller.
Achieve robust soccer robot control by decoupling high-level tactical decisions from low-level motor control using a hierarchical RL framework that transfers well from simulation to the real world.
Controlling soccer robots involves multi-time-scale decision-making, which requires balancing long-term tactical planning and short-term motion execution. Traditional end-to-end reinforcement learning (RL) methods face challenges in complex dynamic environments. This paper proposes HierKick, a vision-guided soccer robot control framework based on dual-frequency hierarchical RL. The framework adopts a hierarchical control architecture featuring a 5 Hz high-level policy that integrates YOLOv8 for real-time detection and selects tasks via a coach model, and a pre-trained 50 Hz low-level controller for precise joint control. Through this architecture, the framework achieves the four steps of approaching, aligning, dribbling, and kicking. Experimental results show that the success rates of this framework are 95.2\% in IsaacGym, 89.8\% in Mujoco, and 80\% in the real world. HierKick provides an effective hierarchical paradigm for robot control in complex environments, extendable to multi-time-scale tasks, with its modular design and skill reuse offering a new path for intelligent robot control.
